DESC:FIELD OF INVENTION
The present invention relates to real-time data capturing and processing system for ongoing qualitative assessment of road characteristics. More specifically, the present invention relates to the field of digital image processing, computer vision, and applied machine learning for identifying numerical score for attributes of roads, in general.

BACKGROUND OF INVENTION
In the domain of road transportation, in the recent times of exponential increase in the count of vehicles present onto the roads, thereby, the density of traffic, the act of driving has become arduous specially when un-disciplined driving is faced by other driving participants. Further, when multiplicity of vehicles are concurrently present in the parallel driveways, it congests the lane of driving which tends to intrude the comfort driving space of driver in the cases of diversions & overtaking.

Aforesaid phenomenon of making the driving lane congested makes difficult to identify discontinuities present over the road for driver, and anticipate forthcoming obstacles/discontinuities etc. Additionally, when driving is performed in varying ambience & illumination conditions, due to the lack of visibility the detrimental effects of unstructured heterogeneous roads become more prominent. Thereby, hikes occurrence of accidents.

As a solution, automated electrical means are provided so as to identify several elements of the road and based on the trained logic provide an intimation for characteristics of road to accordingly advise the drive to select driving parameters and make diversions.

Patent bearing No. US20130293717A1, filed by: GM Global Technology Operations LLC, dated: 09/04/2013, discloses a system and method for providing lane sensing on a vehicle by detecting roadway lane-markers, where the system employs a surround view camera system providing a top-down view image around the vehicle. The method includes detecting left-side and right-side lane boundary lines in the top-down view image, and then determining whether the lane boundary lines in the image are aligned from one image frame to a next image frame and are aligned from image to image in the top-down view image. If the boundary lines are not aligned, then calibration of one or more of the cameras is performed, and if the lines are aligned, then a model fitting process is used to specifically identify the location of the boundary lines on the roadway.

Patent bearing No. US20140003709A1, filed by: , dated: 15/08/2015, discloses a system and method are disclosed for detecting road marking in a video using learned road marking templates. The system comprises a template learning module configured to learn the feature-based road marking templates from a set of training images. The template learning module is configured to rectify each training image, detect multiple regions of interest, and for each detected region of interest, detect multiple key points. The template learning module extracts feature vectors for the detected key points and builds the road marking templates from the feature vectors. The system also includes a road marking detection module for detecting road markings in a video at runtime using the learned road marking templates. During runtime, these templates are matched using a two-step process of first selecting promising feature matches and subsequently performing a structural matching to account for the shape of the road markings.

Patent bearing No. WO2017208264A1, filed by: Ranjeet Deshmukh, dated: 03/06/2016, discloses a method and system for determining road roughness and unevenness using an application. In one embodiment, the method includes installing a handheld device having an application over a vehicle, estimating, Road Quality Index (RQI) value of roughness and unevenness of a pavement surface, based on a plurality of parameters which are generated due to vibrations, GPS location, Speed, tilt Angles, direction of the handheld device during motion of vehicle using a plurality of sensors, mapping the estimated Road Quality Index value to several international standard values to categorize the quality of the road and standardizing the plurality of parameters generated from the respective sensors by identifying the frequency of data reads and calibrating the same.

Patent bearing No. US9108640B2, filed by: Google LLC, dated: 31/01/2012, discloses Systems and methods for monitoring vehicle sensors to determine and report road quality using a communication device are disclosed. The communication device determines the vehicle's location on a road, such as by use of a GPS-enabled head unit or similar device and appropriate mapping software. Monitoring road quality may be achieved by adding a sensor to the shocks, by use of a vertical displacement sensor present on the head unit, and the like. Various combinations of sensors may be employed. A horizontal displacement sensor may be used. The signals from the sensors are monitored by the head unit and analyzed to judge the quality of the road by the amount of vertical vibration that is encountered. This data, together with the vehicle's location, may be transmitted through a mobile network to a central server for distribution in road quality reports and to improve driving directions in mapping software.

However, each of the aforementioned technologies lacks in one or other aspects when compared to present invention, majorly addressing identification lane markings, lane curvature being essential for lane departure warning, identification of road sign even in the occluded scenarios regardless of day or night, provision of temporal timestamp, and identification of material & width of road along with connectivity to other coinciding roads.

Thereby, the general purpose of the present invention is to provide an improved combination of convenience and utility, to include the advantages of the prior art, and to overcome the drawbacks inherent therein.

OBJECTS OF THE PRESENT INVENTION
The principal object of the present invention is to provide a system to facilitate assessment of the attributes of road in real-time basis, to compute a qualitative opinion in terms of a numerical score, for the assistance of driver.

Consistent with the precedent object, further object of the present invention is to provide a method by virtue of which elements of characteristics of road such as Lane characteristics, Road characteristics, presence of Potholes, & other road assets can be assessed, to compute a qualitative opinion in terms of a numerical score, for the assistance of driver.

Other object of the present invention is to provide a system which can provide adaptive mechanism, for dynamic classification & selection of elements of the road characteristics, which is competent to process & distinguish discrete set of high resolution video & pictorial data, may be in the spatial or non-spatial format.

Another object of the present invention is to provide a system wherein probabilistic modelling approach based on pre-trained algorithm can provide confidence score for bounding boxes.

Another object of the present invention is to provide a method wherein Weighting along with Worsening Factor influence overall score of the system in such a manner that, the resultant output of the system become more realistic.

SUMMARY OF THE PRESENT INVENTION
The following presents a simplified summary of the invention in order to provide a basic understanding of some aspects of the invention. this summary is not intended to identify essential features of the claimed subject matter nor it is intended for use in determining or limiting the scope of the claimed subject matter. Its sole purpose is to present some aspect of the invention in a simplified form as a prelude to a more detailed description of the invention presented later.

In one aspect of the present invention, the system meant for assessing attributes of road entails plurality of input interfaces such as sensors for collecting video or pictorial data for surface of road, sub-surface of road, traffic & other obstacles present over the road, and ambient seasonal factors, which is thereafter fed to the control block (500) performing systematic arrangement & sorting of collected data, manipulating in the form of computer readable format, and processing of data based on the trained algorithms equipped with pre-defined logic.

In further aspect of the present invention, while processing the collected data as the part of the control action weights are assigned for each of the input variables within the parent-set of input variables. Which implies, within the parent-set of input data fractional values are supplemented to each of the sub-set input variables for normalizing the influence of the involved variables, by means of assigning weight may be in the form of worsening factor or ranking consideration; for computation of overall score of the present system (10).

In another aspect of the present invention, for determining the overall score of the present system the Lane Characteristics (100) is employed as a one of the parent data set of input variables, which has a sub-set of input variables comprising visibility of lane (110), lane curvature (120), temporal timestamp information (130), and occlusions (140) present over the road.

In another aspect of the present invention, for determining the overall score of the present system the Road Characteristics (200) is employed as a one of the parent data set of input variables, which has a sub-set of input variables comprising number of lanes (210), width of the lane (220), width of the road (230), and type of material utilized (240) for the road construction.

In another aspect of the present invention, for determining the overall score of the present system wherein presence of Potholes (300) is employed as a one of the parent data set of input variables, which involves data related to width of pothole, length of the pothole, depth of pothole, and count of potholes.

In another aspect of the present invention, for determining the overall score of the present system the Other Road Parameters (400) is employed as a one of the parent data set of input variables, which has a sub-set of input variables comprising type/nature of the road (410), traffic lights & signs (420), lightning conditions (430), and seasonal factors (440).

BRIEF DESCRIPTION OF DRAWINGS
The advantages and features of the present invention will become better understood with reference to the following more detailed description taken in conjunction with the accompanying drawings in which:
illustrates flow of instructions for computation of overall Score for Road Quality (SRQ) according to one embodiment of the present invention;
illustrates flow of instructions for computing score of Lane characteristics (100) according to one embodiment of the present invention;
illustrates flow of instructions for computing score of Road characteristics (200) according to one embodiment of the present invention;
illustrates flow of instructions for computing score for potholes (300) according to one embodiment of the present invention;
illustrates flow of instructions for computing score for other road traffic parameters (400) according to one embodiment of the present invention.
Illustrate lane curvature parameter, the curvature is calculated per frame
Provide details on Camera and Hardware used for Tests

Like reference numerals and names refer to like parts throughout the several views of the drawings

DETAILED DESCRIPTION OF THE INVENTION

In the following description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the present invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced without these specific details.
Reference herein to “one embodiment” or “another embodiment” means that a particular feature, structure, or characteristics described in connection with the embodiment can be included in at least one embodiment of the invention. The appearances of the phrase “in one embodiment” in various places in a specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Further, the diagrams representing one or more embodiments of the invention do not inherently indicate any particular order nor imply any limitations in the invention.

As used herein, the term “plurality? refers to the presence of more than one of the referenced item and the terms “a”, “an”, and “at least” do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced item.

It should be emphasized that the term “comprises/comprising” when used in this specification is taken to specify the presence of stated features, integers, steps or components, but does not prelude the presence or addition of one or more other features, integers, steps, components or groups thereof.

The term ‘present system’ is interchangeably used to show presence of the ‘present invention’ in the following description, and relates to essentially the same subject matter.

Fig. 1 to Fig. 5 shows flow of instructions for computing various parameters influencing output of the present invention (10) when measured in terms of numerical integer value, which essentially indicates quality or characteristics or attributes vested with the outer surface of road and obstacles associated with it, whist the system proposed for fulfilling aforesaid functionality is named as “System for Indicating Road Quality Index and Method Thereof” and described hereinafter.

Referring to the Fig. 1, it shows overall architecture employed for performing assessment of numerous attributes of roads in general, to which the present system is likely to be exposed.

Further, the Fig. 1 delineates an interlinked network wherein supplementation for numeric values for correspondingly assigned variable after going through weighting reaches the control block (500) which based upon pre-fed data provides numerical indication for good or bad attributes of involved span of the road, for determination of Safe or Unsafe characteristics of road which in turn will help driver for anticipating experience of drive for travelling/cruising through the said involved road, along with extent to safe or extent to unsafe provided in the scale of 1 to 10 integers accordingly for better assessment of attributes of the road.

The present system of road quality assessment and prevailing Road Quality Index (RQI) indicating essentially intakes and relies upon four parent-set of distinct input variables namely:
First set of input variables defining geometrical contours, homogeneity, and occlusion located onto the road;
Second set of input variables defining molecular arrangement based on material incorporated, and dimensional bifurcation of road;
Third set of input variables defining discontinuities present at the outer surface or sub-surface of road;
Fourth set of input variables defining influence of interlinking connectivity between multiplicity of separate roads, usual density of automobiles, illumination effects, and seasonal disturbances.

Parent sets of input variables as prescribed in above-para are further categorized in plurality of sub-sets having predefined designation for each of the following sub-entries, collectively indicating numerical score of each of respective parent-set of data.

Further, process of weighting is performed for each of the parent set of input data of the present system, particularly at the time of combining inputs of the sub-set of entries for computation of score of corresponding parent-set.

Wherein, the term weighting can be defined as act of converting initial set of data more realistic based on the actual importance/influence of the certain entity over other entries belonging same set of inputs, for ensuring reasonably realistic output of the present system. More specifically, in the process of weighting fractional value between 0 to 1 is assigned to each of the individual entry, in such a manner that summation of all the assigned fractional value invariably comes to be 1.

Referring to the Fig. 2, which illustrates First parent-set of input variables defining geometrical contours, homogeneity, and occlusion located onto the road, essentially relating to the flow of instructions required for computation of the lane characteristics (100) according to one embodiment of the present invention.

Wherein, the term lane can be characterized by segment of road which is divided using thick visible markings for making parallel ways for facilitating transportation, while lane characteristics (100) is the attributes taken into consideration for determining score for the first set of input variables.

The parent-set of Lane Characteristics (100) is categorized into four sub-sets as summarized below, and as can be referred from Fig. 1 & Fig. 2:
P1: Lane Characteristics:
P1.1: Lane Visibility;
P1.2: Lane Curvature Estimation;
P1.3: Temporal Timestamp;
P1.4: Lane Occlusions.

With reference to the first sub-set of input variable P1.1 (110) of the lane characteristics (100), visibility of the lane is ordinarily notable/identifiable presence or absence of edges of lane, which can either be full or partial.

The said element can be determined using various vision and image based techniques such as LaneNet, YOLO etc. maybe with or without incorporation of edge detection (111) algorithms accountable for identification of lane boundaries by checking the continuity in the edges. Such algorithms for identification of boundaries of lanes may include techniques such as but not limited to Canny edge detection, Sobel detection etc.

Further, a threshold is to be defined which in-conformation to the employed algorithm may denote overall visibility of the involved span of lane, as followed:
If (edge continuity) >> threshold ? visibility of lane is good;
If (edge continuity) ~ threshold ? visibility of lane is just ok (acceptable);
If (edge continuity) << threshold ? visibility of lane is bad.

With reference to the second sub-set of input variable P1.2 (120) of the lane characteristics (100), the lane curvature estimation is one of the most influential element for direction and trajectory planning, as road/lanes having extreme blind curves many a times leads to accidents during intersections.

In the present system lane curvature analysis with mathematical modelling is estimated using advanced vision and machine learning algorithms; and incorporates geometric modelling (121) and attention mechanism utilizing advanced deep learning algorithms such as but not limited to Agnostic lane identification, Attention based lane mapping, key point based lane estimation and analysis.

With reference to the third sub-set of input variable P1.3 (130) of the lane characteristics (100), the temporal timestamp is yet another important parameter which provides assistance to drives during lane departures, and provides notion about general attributes of road for prolong span.

Wherein, it is of utmost importance to have nearly homogeneous lane visibility and thickness after every periodical distance, say every 100m or 200m length. Which in turn would not only help in analyzing the condition of lane boundaries for a reasonably long distance, but also would not penalize the road conditions if the lane marking is not clear for a very small distance.

Moreover, for computation of the temporal timestamp score a Spatio-Temporal based data samples which are collected in real-time is used. Which are thereafter being analyzed by means of various computer vision and time-series based sensor modelling (131) techniques, such as but not limited to CNN-LSTM, RNN, GRU etc.

It is to be noted that, in real-time, the data is extracted from various input sensors such as accelerometer, gyroscope, etc. As the information from camera sensors alone may not be sufficient to have complete perception of road traffic and driving. Hence, it is imperative to have multi-sensor information fused to gain more analysis on real-time lane detection, for performing the sensor modelling approach.

With reference to the fourth sub-set of input variable P1.4 (140) of the lane characteristics (100), the occlusion of lane accounts for inability of driver to have clear perception about lane boundary markings, chiefly due to on-road obstacles such as traffic, and varying illumination scenarios.
Further, when it comes to unstructured road traffic with less discipline driving of other traffic participants, it is highly challenging to identify the lane markings. Thus, it is preferable to use techniques such as but not limited to the vanishing point based lane estimation with perspective transformation to identify the lane line markings during the occluded road traffic scenarios, for employing approach of vision perspective transformation (141).

Method for determining score for Lane Characteristics:-
The method includes steps of determining corresponding values of each of the sub-set variables, assigning weights to each of the said variables, and lastly computing overall score for lane characteristics using Weighted Sum Average (WSA), Weighted Multiplicative (WM), or Entropy method.

Criteria applicable for each of the sub-set variables P1.1-P1.4
1) Score for Visibility of Lane (P1.1)
The score for visibility of lane (110) is calculated in linear manner preferentially using linear scale of 1 mm – 100 mm which proportionally correlates to integer value laying between range of 1–10.
For instance, if the measured value of lane thickness is 1mm score for visibility shall be 0, if the measured value of lane thickness is 10 mm score for visibility shall be 1, if the measured value of lane thickness is 20 mm score for visibility shall be 2, if measured value of lane thickness is 100 mm score for visibility shall be 10. Which implies linear interval between measured values in the steps of 10 mm’s.
2) Score for Lane Curvature Estimation (P1.2)
In case of lane curvature estimation (120), distance of curved surface of lane on a road can be considered. For example:
Consider 1km road in which 200 m of road has lane curvature that means it is out of 1000 m road, 200 m of road is curved and 800 m road is not curved (straight line). Here it can be considered that normalizing the value for curvature as 200/1000 = 0.2.

After normalizing the values of lane curvature:
Condition 1: If normalized value is in range 0 – 0.4 that means very little portion of road is curved and can be assumed with a rank in the range of 7 to 10;
Condition 2: If normalized value is in range 0.5 – 0.7 that means more than half of the road is curved and can be assumed with a rank in the range of 4 to 6;
Condition 3: If normalized value is in range 0.7 - 1 that means full road is curved and can be assumed with a rank in the range of 1 to 3.
Consider a road consists of lane line with thickness of lane be 50mm range, count of occlusion of lane n is 5 and temporal timestamp be assumed as 500 samples per 100 meters (ideal value z = 4), curvature of road covered is 200m per 1000m.

The lane curvature parameter, the curvature is calculated per frame for a range of 2.5 to 5 meters based on lane markings. For easy reference, the qualitative results are shown in figure

3) Score for Temporal Timestamp (P1.3)
For determining temporal timestamp (130), exemplary average values are considered for linear case, wherein measurements for 100M road is given by x, and the measurement for 1 meter (calculated value based on overall available samples) would be represented using y = x/100.
For 100M road, if it is assumed that there are say, 5000 measurements, then in the given case, y would be 5000/100 = 50 for 100M distance.
Further, z is an ideal measurement value per meter length, and can be substituted as 40 per meter without considering the average.
Wherein:
Condition 1: y << z, then a score of 1–3 would be assigned based on other parameters such as actual temporal timestamp readings that would vary in real-time;
Condition 2: y ~ z, then a score of 4–7 would be assigned based on other parameters, considering that the temporal time stamp readings are uniform;
Condition 3: y >> z, then the score of 8–10 would be assumed based on other parameters; considering the temporal time stamp readings are better than the ideal case z.

4) Score for Occlusion of the Lane Marking (P1.4)
Inhere, n = count of occluded entities present on the road (i.e. vehicles, road boundaries, garage etc.) is considered for the computation of the lane occlusions (140).

Notably, a tanh function (i.e. Tan hyperbolic function) is best suited for conditions where it is required to represent both the positive and negative extreme cases that would result in a normalized function. The tanh function also helps to reduce the weighting parameters, i.e. instead of weighting the influence factors and their combinations, a final scalable parameter can be given.

A tanh is a function which gives the values in the range of -1 to 1. In the score case, -1 and – 0 cannot be considered, so 0 – 1 is considered. Also 0.96 is considered for extreme case as 0.9 – 1 has similar value in tanh function with varying decimal values. On the other side 0.2 is considered as least case as 0.1 – 0.3 has similar value in tanh function as compared to 0.9 – 1.

In the exemplary case, assuming a score of 0.2 for n < 1 and assuming a score of 0.96 for n > 10 after the application of tanh function, the equation could be written as:
For n > 10: tanh (10*s + b) = 0.96 resulting in
10*s + b = 1.94 ? (1)
For n < 1: tanh (1*s + b) = 0.2 resulting in
1*s + b = 0.202 ? (2)

Solving (1) and (2), one would get s = 0.202 and b = 0.08;
wherein s = scaling parameter, and b = biases.
Furthermore, assuming occlusion count as n = 15,
tanh (15*s +b) = tanh (15*0.202+0.008) = 0.99

Similarly, for occlusion count as n = 2;
tanh (2*s +b) = tanh (2*0.202+0.008) = 0.38
Sample calculation for determining score of the Lane Characteristics:

Exemplary values:
Thickness of Lane = 50mm,
Count of occlusions, n = 5,
Temporal Timestamp being 500 samples per 100m length, (ideal value = 4)

Weights for sub-set of input variables:
P1.1 = 30%, P1.2 = 20%, P1.3 = 25%, P1.4 = 25%
P1.1: as thickness of lane is 50 mm, the score would be 5 based on linear relation.
P1.2: Lane curvature = 200/1000 = 0.2 that means a road consists of very little portion curved area. A rank of 7.5 can be assumed (divided by 10 as 7.5/10 = 0.75 for range 0 – 1)
P1.3: as value of timestamp is 500 samples per 100m distance, then y = 5, and the score would be 8.
P1.4: as value of the occlusion is 5; tanh (5*0.202+0.08) = 0.8 (based on calculated s and b values)
Assuming the ranking technique used as Weighted Sum Average (WSA);
WSA = ?_(i=1)^m¦?xi*wi/i?i

Wherein, xi = elements values, Wi = weights and m = number of elements,
The total score would be:
(0.3*0.5) + (0.75*0.2) + (0.8*0.25) + (0.25*0.8) = 0.7
While the score of lane characteristics on a 10-point scale would be 0.7 * 10 = 7.

Referring to the Fig. 3, which illustrates second parent-set of input variables defining molecular arrangement based on material incorporated, and dimensional bifurcation of road, essentially relating to the flow of instructions for computing score of Road characteristics (200) according to one embodiment of the present invention.

Wherein, the term road characteristics (200) is the attributes taken into consideration for determining score for the second set of input variables, while the parent-set of Road Characteristics (200) is categorized into four sub-sets as summarized below, and as can be referred from Fig. 1 & Fig. 3:
P2: Road Characteristics:
P2.1: Number of Lanes;
P2.2: Width of the Lane;
P2.3: Width of the Road;
P2.4: Type of Road Material.

With reference to the first sub-set of input variable P2.1 (210) of the road characteristics (200), number of lanes are parallel segments of the road and identifying their count is inevitable for issuance of warning or assisting the driver in real-time during lane departures specifically in urban roadways or national/express highways.

Further, the count of lanes can be determined using advanced machine learning algorithms, and computer vision & CNN based techniques such as but not limited to LaneNet, also YOLO could be utilized for localizing the lanes, as a part of vision based transformations (211). After identification of lane markings, count of the detected lines through a number of bounding boxes or a number of lines that are recognized as lanes.

With reference to the second sub-set of input variable P2.2 (220) of the road characteristics (200), width of the lane is the linear distance between opposite edges of the lane of two consecutive lane markings, and can be determined by measuring width of the bounding boxes or lines, or else opposite edges of lane. Measuring width of lane as a part of lane geometry (221) approach is expressed as bellow, wherein width of lane can be scaled according to the requirement:
Width of lane = |x2-x1|
Where, x2 and x1 are horizontal coordinates of bounding boxes or lines.

With reference to the third sub-set of input variable P2.3 (230) of the road characteristics (201), width of the road typically varies with respect to different nature of roads such as urban, highway traffic. Using segmentation the road can be detected in the image and width of the segmented road can accordingly be calculated, followed scaling for determining the actual width of the road, using vision (231) based techniques.

Wherein, segmentation is a task of masking the exact spatial information of the image, for getting accurate pixel wise area, length and width which can be scaled as per the requirement. The method for calculating width of the road is expressed as below:
Width of Road = |x2-x1|
Where, x2 and x1 are horizontal coordinates which can be taken from segmented mask of the road.

With reference to the fourth sub-set of input variable P2.4 (240) of the road characteristics (200), material of the road can be classified into tar, concrete, gravel, earthen, while driver’s drivability depends on type of road material.

Further, type of material used in the road can be identified using techniques such as EfficientNet, VGG, Inception, and said algorithms use the deep Convolutional Neural Networks (CNN) in different sizes and combinations of layers, following multi-classification approach (241).

Method for determining score for Road Characteristics:-
The method includes steps of determining corresponding values of each of the sub-set variables, assigning weights to each of the said variables, and lastly computing overall score for road characteristics (200) using Weighted Sum Average (WSA), WM, or Entropy method.
Criteria applicable for each of the sub-set variables P2.1 - P2.4
1) Score for Number of Lanes (P2.1)
It could be considered that, number of lanes (210) for local urban roads would be in the range of 1 or 2, and may increase in the cases of national or express highways. Hence, the score range based on the number of lanes would be as follows:
0 (Zero) Lane - Minimum score of 1.
1 to 2 Lanes - Score range of 2 – 4
3 to 4 Lanes - Score range of 5 – 6
More than 4 - Score range of 7 – 10
2) Score for Width of Lane (P2.2)
The standard width of lane (220) is typically 1m, i.e., 1000 mm. Thus, if the actual lane width is 700 mm, the normalized converted width shall be 0.7, thereby, the score is assumed to alter linearly with the increasing or decreasing lane width.
3) Score for Width of Road (P2.3)
The standard width of road (230) is 3.75 meters, further, in broad highways the width of the road can be higher said value. Thus, consistent with aforesaid values, the score range based on width is as follows:
1 to 2 meters - Score range of 1 – 2
2 to 3 meters - Score range of 3 – 5
3.75 meters - Score range of 6 – 8
> 3.75 meter - Score range of 8 – 10
It is to be noted that the score range can overlap across different width size.
4) Score for Road Material (P2.4)
For construction of road, Tar or Concrete is preferred over Gravel and Earthen, as a type of road material (240). Hence, it is considered that for Tar or Concrete road would have a better score than Gravel and Earthen road (i.e., numerically, Tar or Concrete road score would be one or two units higher), provided all other conditions remain the same.

Sample calculation for determining score of the Road Characteristics:
Exemplary values:
Number of Lane = 4,
Width of Lane = 60 mm,
Width of Road = 5M,
Type of Material = Tar.
Weights for sub-set of input variables:
P1.1 = 30%, P1.2 = 20%, P1.3 = 25%, P1.4 = 25%
P2.1: given that there are 4 lanes, the score would be 7 and normalized value being 0.7.
P2.2: given that the width is 60 mm, the normalized score would be 60/100 = 0.6.
P2.3: since the width of road is > 3.75, the minimum score would be 9/10, i.e., 0.9.
P2.4: since a Tar road is considered, the score is taken as 0.8
Assuming the ranking technique as Weighted Sum Average (WSA),
WSA = ?_(i=1)^m¦?xi*wi/i?I
wherein xi = elements values, wi = weights and m = number of elements;
The total score would be:
(0.3*0.7) + (0.6*0.2) + (0.8*0.25) + (0.25*0.8) = 0.73
Hence, the score for road quality is 0.7*10 = 7.3.

Referring to the Fig. 4, which illustrates Third set of input variables defining discontinuities present at the outer surface or sub-surface of road, essentially relating to the flow of instructions for computing score for potholes (300) according to one embodiment of the present invention.
Wherein, the presence and dimensions of potholes is the attributes taken into consideration for determining score for the third set of input variables, while the parent-set of pothole based RQI comprises steps as can be referred from Fig. 1 & Fig. 4.

The term pothole in general can be defined as recess or discontinuity present apparently present onto the external surface of the road, and in the present system attempts are made to determine length, width, & depth of the pothole, alongside overall count (352) of the pothole using count mechanism (350), followed by calculating the worsening factor (360) and score of the pothole (390).

Spatial data for potholes is captured in the form of images and videos considering all the possible vision based real-time road traffic scenarios, the captured video stream is processed into input frames using OpenCV. Then images are resized into their respective shapes as required by the deep learning networks and are normalized between 0 and 1.

In the present system the potholes are firstly detected and then segmented, wherein semantic segmentation (320) is performed for masking the exact spatial information of the image, which helps in getting accurate pixel wise area, length and width which can be scaled accordingly. The existing models which are pre trained on large segmentation datasets like the COCO, VOC, etc. the said datasets can be fine-tuned to segment potholes. Which includes but not limited to Mask RCNN, U-Net, Deeplabv3, etc.

The above mentioned databases however mandate large set of input data for reasonably decent results, which is a tedious task as labelling images for segmentation is highly time consuming. But, there are plurality of alternative shot learning techniques are available for fulfillment of the task, which are very general and can be integrated into almost any network, addressing use of the bounding boxes (330) output with the help of algorithms such as but not limited to GrabCut, DeepCut, DeepGrabCut, & Watershed. The outputs of this branch are the area, width and length parameters of each pothole from each frame.

Detecting the potholes or alike using bounding boxes is a well explored area, where starting from the simple sliding window approaches such as “OverFeat” with the use of “Objects as Point” and the compound scaling methods. The evolution of these algorithms led to an increased accuracy and speed of detections. The present system mainly focuses on recent state of the art techniques such as but not limited to EfficientDet, YoloV4 (scaled) and YoloV5. In which, EfficientDet works on the recent FPN (Feature Pyramidal Networks) and compound scaling which led to both increase and accuracy and efficiency in computation, whereas scaled YOLOv4 is competent with EfficientDet which uses additional cross stage scaling.

The networks mostly used are pre-trained on the VOC and PASCAL datasets and are then fine-tuned on the potholes dataset. The output of this phase is the bounding boxes (330) for each pothole and for each frame.

Apart from that, while determining size of the potholes, depth estimation is often a one of the challenging problems in computer vision, when data collection is solely dependent on conventional sensors like accelerometer and gyroscope, and advanced 3D sensors like LIDAR, thermal images, and other sequential imaging methods are not duly incorporated. Generating a final heat map that gives the relative depth estimation (340) using inverse depth transformation is found to be highly favourable for the present system, which employs adversarial networks like CycleGAN, Pix2Pix or other segmentation networks like depth CNN, pose CNN for generating the heat map from a single image, which have achieved significant success.

Further, Pothole Count (352) is another important parameter which plays influential role in the present system. There are many counting algorithms available but not limited to simple centroid based trackers, and complicated (Kalman filter + deep visual features) trackers. The said complicated trackers may include but not limited to DeepSORT, FairMOT, Siamese based Networks, etc. Wherein, in the case of present system a simple centroid based tracking with proper parameter adjustments is proficient for counting the potholes, as unlike other scenarios where the objects have their own independent motion potholes are stationary and only the camera moves. This makes all the potholes move in the same spatial distance and direction. Thereby, tracking them with their centres works in par when compared with most of the complicated trackers.

Further for the Pothole Count (352) the camera and hardware configuration:
CPU : SnapDragon 720g, Octacore, 2.3GHz, 8nm
RAM: 6GB
Camera : HM2 Sensor, 108MP, f/1.88 Aperture, 6P lens, 1/1.52” sensor size
OS: Android 11
App Size: 63 MB
Inference Time (avg) : 600 ms
Figure 7 Provide details on Camera and Hardware used for Tests

Worsening Factor [WF]:-
The term Worsening Factor (360) is a numerical criterion indicating extent to which a pothole is deleterious depending upon corresponding depth, width, & length of the involved pothole, and measured in linear integer scale of 1–10.
One of the simplest methods for computing the Worsening Factor (360) is additive weighting, which assigns fractional weight to each factor such that the result sums to 1 (or 10 for scaled version). The effect of each parameter is given to tanh function to determine the output between 0 and 1(multiply 10 for scaling), and the exact scaling by the tanh function is expressed as:
WF = w1 * ( wph/Wroad ) + w2 * tanh( dph x Ssp1 + b1) + w3 * tanh( lph x Ssp2+ b2)
where, wph is the width of the pothole; Wroad is the width of the road, dph is the depth of the pothole, lph is the length of the pothole; W1, W2, W3 are weightages for width, depth and length of the potholes, and Ssp1, Ssp2 - scaling factor for sub-parameters to be determined by the data and manual examination and b1, b2 are biases.

It is to be noted that, width and depth of the pothole have higher importance, hence the weighting factor for width (W1) and depth (W2) would be higher than that of length (W3).
Sample calculation for determining score of the Potholes:
Exemplary values:
Length of pothole, lph = 5cm,
Width of Road, Wroad = 3m (300cm),
Pothole Width wph = 5cm,
Weights for sub-set of input variables:
Width of pothole, W1 = 04,
Depth of pothole, W2 = 04,
Length of pothole, W3 = 0.3,

Scaling Factor & biases Calculation:
For Depth: Assuming the worst depth is 5cm, corresponding RQI for depth would be considered as 0.95; Similarly, for depth is 1cm, corresponding RQI for depth is 0.2. Thereby:
tanh (5*S1*b1) = 0.95
Given that, 5*S1+b1 = tan-1 (0.95),
tanh (1*S1*b1) = 0.2, and
[1*S1 + b1] = tan-1 (0.2)
It can be found that S1 = 0.40725 cm-1 and b1 = -0.20455
For Length: Assuming the worst length is 10cm, corresponding RQI for length is 0.95. Similarity for length of 1 cm, the RQI for depth is 0.1.
Thereby:
tanh (5*S2*b2) = 0.95 and 5*S1+b2 = tan-1 (0.95)
Therefore;
tanh (1*S2*b2) = 0.1 implying 1*S2+b2 = tan-1 (0.1)
Solving, one would get:
S1 = 0.1924 cm-1 and b1 = -0.092
Worsening Factor Calculation:
The worsening factor can be computed as:
= 0.4 * (5cm / 300cm) + 0.4 tanh (3*S1+b1) + 0.3 tanh (5*S2+b2)
= (2 / 300) + 0.4 * 0.768 + 0.3 * 0.7013.
= 0.5243.

Inference: Given a pothole length of 5 cm, width of 5 cm and depth 3 cm in a road of wide 300 cm, the worsening factor is 0.5243.

Wherein, the weighting method as expressed above, the ranking can be fixed manually or can be estimated using AHP (Analytic Hierarchical Process) which is highly intuitive and solid way of setting the weights to the above parameters; which in turn suggests a hybrid ranking (such as AHP) and a simple additive weighting mechanism.

To calculate the RQI for potholes (390), individual intensity factors should be considered and another important thing is that consecutiveness and the closeness of two or more potholes make the conditions worse.
For instance, if N is the number of potholes,
F = WF1 + WF2 + …. + WFN + WF1,2 + WF1,3 + …… + WFN,N-1
P = w2 x tanh( F x S)
Where, WFi,j is the worsening factor of combined ith and jth pothole, and it is expressed as:
WFi,j = (WFi + WFj) x ( c1 / wi,j + c2 / li,j)
Where c1 > c2; wi,j is width-wise distance between potholes and li,j is the length-wise distance between potholes. Whereas, S is the final scaling parameter to be determined by some manual examination of data.

Referring to the Fig. 5, which illustrates Fourth parent-set of input variables defining influence of interlinking connectivity between multiplicity of separate roads, usual density of automobiles, illumination effects, and seasonal disturbances, essentially relating to the flow of instructions for computing score for other road traffic parameters (400) according to one embodiment of the present invention.

Wherein, the term other road traffic parameters are the attributes taken into consideration for determining score for the fourth set of input variables, while the parent-set of Other Road Parameters (400) is categorized into four sub-sets as summarized below, and as can be referred from Fig. 1 & Fig. 5:
P4: Other Road Parameters:
P4.1: Nature of road;
P4.2: Traffic lights, & signs;
P4.3: Lightning conditions;
P4.4: Seasonal factor.

With reference to the first sub-set of input variable P4.1 (410) of the other road parameters (400), wherein the nature of road (410) can be classified into national, state, urban, residential, terrain, & mountain roads, and can be further subdivided into General roads (i.e. national, state, urban, residential) and Hilly Roads (i.e. terrain, Mountain roads) based on travelling perspective.
When comes to identification of type/nature of road one of the approach directs use of state of the art classification techniques like EfficientNet, VGG, or Inception, which uses Convolutional Neural Networks (CNN) in different sizes and combinations of layers, and requires more dataset for better classification. While alternate approach directs to use segmentation which extracts the road part of the image through a mask and identifies the class of the road.

With reference to the second sub-set of input variable P 4.2 (420) of the other road parameters (400), wherein recognizing the traffic lights and sign boards (420) while crossing through highways or road in general is one of the inherent functionality of the present system, in order to dynamically assist the in-vehicle driver before collision during intersections. The advanced CNN based methods are modelled & implemented as two stream manner, and evaluated for traffic lights detection and sign board recognition in the present system.

With reference to the third sub-set of input variable P4.3 (430) of the other road parameters (400), wherein in the case when driving is performed in night time, the driver usually confronts multitudes of steering illuminations and high beam reflections from opposite lane vehicles, which causes discomfort to the driver and makes driving highly challenging. As a solution, in this SRQ, the advanced GANs and image super resolution methods is used in the present system for visual perception of road traffic.

With reference to the fourth sub-set of input variable P4.4 (440) of the other road parameters (400), wherein the driving patterns varies in day light, night time and is complex especially in rainy, snowy and foggy weather conditions. Thus in the present system, in SRQ, various image corrections methods with adaptive filtering is employed for better understanding of the road with change in scenarios, to determine the prevailing influence of seasonal factors (440).

Vision and Perspective Transformations (450):
The vision based techniques such as CNN based architectures for traffic sign recognition, super image resolution and image enhancement techniques from histogram equalization to GAN’s could be utilized in the present system to understand the road traffic environment under different climatic conditions.

Method for determining score for Other Road Parameters:-
Weight & Ranking:
Weights (assumed 10% - 25%) can be given as input to the ranking block (460), maybe by the user as input based on requirement. Once all the sub parameters have been calculated and normalized in the same range, then the different approaches can be selected based on the required criteria for overall score of P4 (480). Once the final score is calculated, the score is sent to its successive stage.
Calculation of score for P4:
The spatial information is captured in the form of images and videos considering all the possible vision based real-time road traffic scenarios. Then video stream is processed into input frames using OpenCV, further, images are resized into their respective shapes as required by the deep learning networks and are normalized between 0 and 1, by assigning weights (470) to the parameters. In continuation with above [P1], [P2], [P3], [P4]. The scores of these parameters will be extracted using the ranking approaches such as WSA, WM, entropy, etc.
Once these parameters are extracted, they are sent to the ranking block (460) based on criteria.
Criteria 1:
Based on the importance of parameters, weights are given
If (weights = fixed) ? SMART, AHP-TOPSIS
If (weights != fixed) ? AHP, CRITIC
Criteria 2:
If all the individual outputs of parameters have the same range (for example, 9 for all the parameters), then it is important to analyse deeply for giving an overall score. Otherwise Simple algorithms can be used to calculate the overall score.
If (All the parameters has same range) ? AHP, CRITIC, AHP-TOPSIS
If (All the parameters are not in same range) ? SMART, WASPAS

Once the final score is calculated, the RQI is provided based on the score.

Steps of performing computation of the score of the road with reference to the present invention, as illustrated by Fig. 1 being consistent with inputs of Fig. 2 to Fig. 5, starts with collecting of data using video & pictorial input devices such as sensors for each of the elements of the sub-sets as prescribed in posterior disclosure for the parent-sets namely P1, P2, P3 & P4. Then normalizing input data assigned with each of the sub-sets using weighting or ranking approach for the determination of score of each of the parent-set. Followed by providing opinion for safe or unsafe in the form of numerical integer value ranging in the order of 1–10, as issued by the control block, being accountable for governing overall analytical activity of the present system.

Notably, the individual score is computed for each parent-set (P1 - P4) using approaches such as WSA, WM and for the final RQI, the ranking approaches such as SMART, CRITIC approaches are used.

In one of the preferred embodiment of the present invention, all the subparameter’s individual scores are calculated separately as described above. Once the score of each subparameter is calculated, then the scores are integrated using the methods such as the weighted sum average method and weighted multiplicative methods for driving scenarios - highways, urban, and rural. Same for the case with pothole analysis, we consider the individual sub-parameters score and compute the final worsening factor for providing the overall score of road quality.

In one of the preferred embodiment of the present invention, the present system incorporates adaptive mechanism, which collects & processes data in real-time for originating output in terms of numerical integer value for qualitative analysis of road attributes.

In one of the further preferred embodiment of the present invention, a dynamic classification approach is utilized as pre-processing step for detecting/identifying the elements/objects of the road, wherein a rigorous examination is conducted for identifying and then grouping the objects present at the surface or sub-surface of road. Further, as a part of enhancement technique dynamic classification is being accompanied with the dynamic selection of the objects.

In one of the further preferred embodiment of the present invention, a probabilistic modelling is used which based on the trained data using plurality of probabilistic & deterministic algorithms provides confidence score for bounding box.

In one of the further preferred embodiment of the present invention, multi-modal information is attained in the forms of qualitative numerical value with consideration of worsening factor accounting for parameters like lane boundaries, road material, potholes, width of the road, for assessing attributes of road.

The minimum and maximum ranges for each parameter are given below in detail with respect to different cases of Road Quality - Good, Average and Bad cases. A default score is also given to each parameter if there are no lane markings.

Lane Visibility and Thickness
For visibility of lanes, continuity of indices (x values) in x-axis of the histogram can be counted.
If the count of missing indices is less than 10, the visibility is marked as good (8 to 10).
If the count of missing indices is from 11 to 50, the visibility is marked as average (5 to 7).
For the remaining cases, the visibility is marked as bad (1 to 4).

Range of Visibility:
Highways and Urban: Good visibility is in range 8 to 10, average is in range 5 to 7 and bad is in range 1 to 4.
Rural: Good is (6 to 10), Average is (4 to 5), Bad is (1 to 3)

For the thickness of lanes, the count of intensity values in the y-axis is considered.
If the count of intensity values (y values) greater than 200 are more compared to other intensities, then thickness is best (8 to 10). If the number is between 100 and 200, then the thickness is good (5 to7) .
For remaining cases, the thickness is considered bad and in the range of 1 to 4.

Range of Thickness :
Highways - Good - (8 to 10) , Average - (5 to 7), Bad - (1 to 4)
Urban - Good - (8 to 10) , Average - (5 to 7), Bad - (1 to 4)
Rural - Good - (6 to 10) , Average - (4 to 5) , Bad - (1 to 3)

No lane case - For both thickness and visibility, the minimum and maximum score would be one.

Lane Curvature
Range:
If a change in curvature varies more than is variation is in the range of more than 3000m, the score is given less (1 to 3)
If a change in curvature varies medium that is variation is in the range of more than 500m and less than 3000, the score is given less (4 to 6)
If a change in curvature varies medium that is variation is in the range of less than 500m, the score is given less (7 to 10)
No lane case: If lane curvature is not possible due to lack of lanes, then the default score considered is 2 to 4.

Temporal Timestamp
Here 50 meters of highway scenario is considered and an average of all values for lane visibility, lane thickness, lane width, road width is taken.

Range:
If the deviation (difference) between average and original values is less than 1, the score would be 8 to 10.
If the deviation (difference) between average and original is a range of 2 to 4, then the score would be 5 to 7 and
In the remaining (difference) case where deviation would be greater than 4, the score would be 1 to 4.
Default case: If there is no temporal timestamp, the default score would be in the range of 5 to 7 (assuming the condition to similar across the entire road)

Lane Occlusion
Range:
No occlusions mean score would be 8 to 10.
Occlusions are less than 50 (count of obstacles) then score would be 5 to 7.
Occlusions are more than 50 (count of obstacles) then the score would be 1 to 4.
For current analysis, no occlusion scenario is considered with a default score of 8 to 10.

Number of lanes

The number of lanes considered for the analysis is two-lane roads.
Range:
The min and max score is given for No lane case 0 lanes - 1 score
1 lane ? (2 to 4) score
2 lanes ? (5 to 6) score
3 or 4 lanes ? (7 to 10) score

Lane Width
The standard lane width in India is 3.75m wide.
If the deviation (difference) between calculated lane width and standard lane width is less than 0.5m, then a score would be given in the range of 8 to 10.
If the deviation(difference) between calculated lane width and standard lane width is greater than 0.5m and less than 1m, then a score would be given in the range of 5 to 7.
If the deviation(difference) between calculated lane width and standard lane width is greater than 1m, then a score would be given in the range of 1 to 4.
NO lane case If there are no lanes, then lane width is considered to be the default score in range 1 to 3.

Road Width
The standard road width in India for a two-lane highway road is between 8m and 10m wide approximately.
The standard road width in India for a four-lane highway road is between 20m and 30m wide approximately.
If the deviation/ difference b/w calculated road width & standard road width is less than 1m, a score is given in a range of 8 to 10.
If the deviation/ difference b/w calculated road width & standard road width is more than 1m and less than 10m, a score is given in a range of 5 to 7
If the deviation (difference) between calculated road width and standard road width is greater than 10m, a score is given in a range of of 1 to 3.

NO lane case: If there is no lane width then for urban areas road width score is considered to default in the range of 5 to 7. For rural areas, the default road width considered is 2 to 4.

Road Material Classification
If the material for the road is tar, then the score would be 8 to 10.
If the material for the road is concrete, then the score would be 8 to 10.
If the material for the road is earthen, then the score would be 4 to 7.
If the material for the road is gravel, then the score would be 3 to 5.

Detailed analysis and comparison showing accuracy levels of observed data obtained with weighted sum and weighted product methods calculation are given in the table below:

Driving Scenario Lane - P1
WSA Lane - P1
WPM Road - P1
WSA Road - P2
WPM
Highway (with lane) 8 8 8 8
Urban (with lane) 8 8 7 7
Urban (without lane) 4 3 5 4
Rural (with lane) 8 8 7 7
Rural (without lane) 5 4 3 3
The above table shows the observed data for different cases of highways, rural, and urban roads. We consider that on highways, there are always lane markings available.
(For more details, please refer to results and figures in pages 32 to 36 in the patent document)

Although a particular exemplary embodiment of the invention has been disclosed in detail for illustrative purposes, it will be recognized to those skilled in the art that variations or modifications of the disclosed invention, including the rearrangement in the configurations of the parts, changes in steps and their sequences may be possible. Accordingly, the invention is intended to embrace all such alternatives, modifications and variations as may fall within the spirit and scope of the present invention.

The foregoing descriptions of specific embodiments of the present invention have been presented for purposes of illustration and description. They are not intended to be exhaustive or to limit the invention to the precise forms disclosed, and obviously many modifications and variations are possible in light of the above teaching.

,CLAIMS:We Claim:

1. A method for indicating road quality for assisting a driver of a vehicle comprises steps of:
a) generating data from image capture device and sensors ;
b) a processing device configured to receive a sensor data and images of an area in a vicinity of a user vehicle captured by an image capture device; collect data related to temporal timestamp, occlusion, count of pothole, area of pothole, and curvature of lane;
c) determine a predicted path of the user vehicle based on analysis of the of images;
d) determining curvature on road indicating a course of the bend in the road by the system;
e) determining road surface in front of the vehicle in a forward direction of travel by the system;
f) determining road condition by pothole and area of pothole by data and images that indicates a condition of a road;
g) determining the road sign and generating road sign information regarding a road sign detected in an environment of the road from the sensor data
h) determining the lane on the road generating lane course information such as lane marking, occlusion of the subject vehicle's lane from the sensor data;
wherein, collect data related to connectivity of selected route, material of road, width of road, number of lanes, temporal timestamp, ordinal occlusion, presence of signals, lightning conditions, actual visibility of lane, count of pothole, area of pothole, seasonal factors, and curvature of lane in the from the image and sensor data are processed and send the same to; assistance module of controller
the data collected by processing device determine temporal timestamp, occlusion, count of pothole, area of pothole, and curvature of lane;
processing device estimated Road Quality Index value and categorize the quality of the road send data for the assistance of a driver.

2. A method for indicating road quality index is comprises steps of:
a) collect data related to connectivity of selected route, material of road, width of road, number of lanes, temporal timestamp, occlusion, presence of signals, lightning conditions, visibility of lane, count of pothole, area of pothole, seasonal factors, and curvature of lane;
b) send data related to temporal timestamp, occlusion coefficient, visibility of lane, count of pothole, area of pothole, and curvature of lane to the processing device;
c) processing device determine temporal timestamp, ordinal occlusion, visibility of lane, count of pothole, area of pothole, and curvature and determine weight of the individual attribute of road;
d) send processed data related to weight of the individual attribute material of road, width of road, number of lanes, temporal timestamp, occlusion coefficient, presence of signals, lightning conditions, visibility of lane, count of pothole, area of pothole, seasonal factors, and curvature of lane for the assistance of a drive;
e) processed data related to weight of the individual attribute of road quality index is processed to communication module connected with controller and server module for indicating attributes of selected road from available possible routes for alerts and notifications of various road conditions and driving conditions.

3. The method for determining road quality index as claimed in claim 2, wherein determining set of road attributes from the data collected from image capture device with data of vehicle sensors, for determining overall road quality index and communication module for interaction with driver, and for assistance of the driver.

4. The method for determining road quality index as claimed in claim 2, wherein method of assigning weights assigned to processing device of the controller for determining attributes of road from attributes measured through data comprises steps of:
a) determine overall accuracy required for computation of set of attributes for specified end point of the selected route;
b) determine individual accuracy required for temporal timestamp, occlusion, count of pothole, area of pothole, and curvature of lane;
c) determine count of data set required to receive from server module collected from image capture device for performing iteration;
d) send set of data collected from server module to the processing device;
e) processing device determine individual variance form the set of data collected timestamp, occlusion, count of pothole, area of pothole, and curvature of lane;
f) assign determined variance to the timestamp, occlusion, count of pothole, area of pothole, and curvature of lane for determination of ordinal temporal timestamp, ordinal occlusion, ordinal count of pothole, ordinal area of pothole, and ordinal curvature of lane;
g) send data related to ordinal temporal timestamp, ordinal occlusion, ordinal count of pothole, ordinal area of pothole, and ordinal curvature of lane to the second computation block of the controller.

5. The method for determining road quality index as claimed in claim 1, wherein image capture device is selected from a camera connected with a motored vehicle, and a camera of a mobile device connected with a motored vehicle.

6. The method for determining road quality index as claimed in claim 1, wherein server module is selected from motored vehicle mounted local server connected with global server, motored vehicle mounted global server connected with central global server, motored vehicle mounted local server connected with central local server further connected with a central global server.

7. The method for determining road quality index as claimed in claim 1, wherein plurality of sensors of vehicle are ultrasonic and depth sensors, accelerometer sensor and gyroscope sensor.

8. A System for determining road quality index comprises the application is initiated and the data's collected from image capture device and sensors determine the score of the Lane Characteristics (100), determine score of the Road Characteristics (200), determine score of the Potholes (300), determine score Other Road Parameters (400), which are processed in parallel with the stored in database and calculate road quality index to display in the handheld device and also transmits in real-time data to storage device.

9. The System for determining road quality index as claimed in claim 8, wherein storage device are such as another computing device, or remote server or cloud server.

10. The System for determining road quality index as claimed in claim 8, wherein determining the score of the Lane Characteristics (100) comprising data of sub-set of input variables selected from visibility of lane (110), lane curvature (120), temporal timestamp information (130), and occlusions (140) present over the road.

11. The System for determining road quality index as claimed in claim 8, wherein determining score of the Road Characteristics (200) comprising data of sub-set of input variables selected from number of lanes (210), width of the lane (220), width of the road (230), and type of material utilized (240) for the road construction.

12. The System for determining road quality index as claimed in claim 8, wherein determining score of the Potholes (300) comprising data of sub-set of input variables selected from width of pothole, length of the pothole, depth of pothole, and count of potholes.

13. The System for determining road quality index as claimed in claim 8, wherein determining score Other Road Parameters (400) comprising data of sub-set of input variables selected from nature of the road (410), traffic lights & signs (420), lightning conditions (430), and seasonal factors (440).

Dated this 17th Day of Nov, 2021.

To Controller of Patents,
The Patent Office,
At Mumbai.