Claims:WE CLAIM
1. A method (300) of detecting a mechanical fault in an autonomous ground vehicle (AGV) (105), the method (300) comprising:
for each of a set of trajectory plan segments along a base path during real-time navigation of the AGV,
receiving (304), by a fault detection device (101), a plurality of vehicle displacement parameters along a given trajectory plan segment; and
determining (306), by the fault detection device (101), an optimal velocity twist of the AGV in the given trajectory plan segment using an artificial intelligence (AI) model (601), based on the plurality of vehicle displacement parameters and a weight of the AGV; and
determining (307), by the fault detection device (101), the mechanical fault in the AGV (105) based on a comparison of an actual velocity twist of the AGV in the given trajectory plan segment and the optimal velocity twist of the AGV in the given trajectory plan segment for each of the set of trajectory plan segments.

2. The method (300) of claim 1, further comprising:
determining (301) the base path based on an initial location of the AGV and a destination location received from a user;
dynamically determining (302) a trajectory plan for a trajectory plan segment along the base path based on a visible navigation region from a current location of the AGV, wherein the trajectory plan comprises velocity-position plan for the AGV (105); and
determining (303) the plurality of vehicle displacement parameters for the trajectory plan segment based on the trajectory plan for the trajectory plan segment.

3. The method (300) of claim 1, wherein the plurality of vehicle displacement parameters comprise an approaching velocity twist of the AGV for the given trajectory plan segment, a departing velocity twist of the AGV for the given trajectory plan segment, an approaching orientation of the AGV in the given trajectory plan segment and an orientation shift in the given trajectory plan segment.

4. The method (300) of claim 1, wherein a velocity twist of the AGV comprises an angular velocity of the AGV and a linear velocity of the AGV.

5. The method (300) of claim 1, wherein the AI model (601) comprises an artificial neural network (ANN) based model.

6. The method (300) of claim 1, further comprising training (305) the AI model (601) using training data to learn a function for determining the optimal velocity twist of the AGV in each of a plurality of trajectory plan segments based on the weight of the AGV, wherein the training data is generated by manually driving the AGV along a predefined path at a predefined speed.

7. The method (300) of claim 1, wherein determining the mechanical fault in the AGV further comprises at least one of:
determining (310) an angular shift of the AGV over the set of trajectory plan segments; and
determining (311) a need for an additional effort required by the AGV to compensate for the angular shift.

8. The method (300) of claim 1, further comprising at least one of:
notifying (308) a user with respect to the mechanical fault; and
compensating (309) for the mechanical fault by initiating a corrective action.

9. The method (300) of claim 8, wherein initiating the corrective action comprises:
from a deviated position of the AGV, suspending (313) dynamic determination of a trajectory plan;
at the deviated position of the AGV and at each new position of the AGV,
determining (314) a direction of a next trajectory waypoint on an originally planned trajectory;
calculating (315) an angle of deviation of the AGV between a current orientation of the AGV and the direction of the next trajectory waypoint;
comparing (316) the angle of deviation of the AGV with a pre-defined threshold value; and
initiating (317) the correction action based on the comparison and the angle of deviation of the AGV.

10. A system (100) for detecting a mechanical fault in an autonomous ground vehicle (AGV) (105), the system (100) comprising:
a fault detection device (101) comprising at least one processor (102) and a computer-readable medium (103) storing instructions that, when executed by the at least one processor (102), cause the at least one processor (102) to perform operations comprising:
for each of a set of trajectory plan segments along a base path during real-time navigation of the AGV,
receiving (304) a plurality of vehicle displacement parameters along a given trajectory plan segment; and
determining (306) an optimal velocity twist of the AGV in the given trajectory plan segment using an artificial intelligence (AI) model (601), based on the plurality of vehicle displacement parameters and a weight of the AGV; and
determining (307) the mechanical fault in the AGV (105) based on a comparison of an actual velocity twist of the AGV in the given trajectory plan segment and the optimal velocity twist of the AGV in the given trajectory plan segment for each of the set of trajectory plan segments.

Dated this 14th day of November, 2019

Madhusudan S T
Of K&S Partners
Agent for the Applicant
IN/PA-1297
, Description:TECHNICAL FIELD
[001] This disclosure relates generally to autonomous ground vehicle (AGV), and more particularly to method and system for detecting mechanical faults in an AGV.