Patent
Techniques to automatically verify object detection, classification, and depth for automated driving systems
| العنوان: | Techniques to automatically verify object detection, classification, and depth for automated driving systems |
|---|---|
| Patent Number: | 11972,612 |
| تاريخ النشر: | April 30, 2024 |
| Appl. No: | 17/178865 |
| Application Filed: | February 18, 2021 |
| مستخلص: | An object detection and classification verification system for a vehicle includes a projection system configured to project a three-dimensional (3D) scene pre-captured at a known distance and comprising at least one known object onto a surface in front of the vehicle a controller configured to verify a performance of an object detection and classification routine by performing the object detection and classification routine on the projected 3D scene to generate a set of results, comparing the set of results to a set of expected results associated with the projected 3D scene, and based on the comparing, determining whether the performance of the object detection and classification routine satisfies a predetermined threshold metric. |
| Inventors: | Li, Dalong (Troy, MI, US); Garbacik, Neil (Lake Orion, MI, US) |
| Assignees: | FCA US LLC (Auburn Hills, MI, US) |
| Claim: | 1. An object detection and classification verification system for a vehicle, the verification system comprising: a projection system configured to project visible light onto a wall surface in front of the vehicle, wherein the wall surface is substantially perpendicular to a ground surface beneath the vehicle; and a controller of the vehicle in communication with the projection system, the controller being configured to verify a performance of an object detection and classification routine by: commanding the projection system to project one of a plurality of three-dimensional (3D) scenes onto the wall surface in front of the vehicle, wherein each of the plurality of 3D scenes is pre-captured at a known distance and comprises at least one known object; performing the object detection and classification routine on the projected 3D scene to generate a set of results; comparing the set of results to a set of expected results associated with the projected 3D scene; and based on the comparing, determining whether the performance of the object detection and classification routine satisfies a predetermined threshold metric. |
| Claim: | 2. The verification system of claim 1 , wherein the set of results comprises a detected location and detected class for each object in the projected 3D scene, and wherein the set of expected results comprises a known location and known class for each object in the 3D scene. |
| Claim: | 3. The verification system of claim 2 , wherein each of the plurality of 3D scenes further comprises a two-dimensional (2D) box having a size corresponding to the known distance. |
| Claim: | 4. The verification system of claim 3 , wherein the controller is further configured to at least one of: output instructions to move the vehicle towards or away from the wall surface until the size of the 2D box in the projected 3D scene matches the size corresponding to the known distance; and output instructions to move the vehicle left or right relative to the wall surface until there is no projection distortion of the 2D box in the projected 3D scene. |
| Claim: | 5. The verification system of claim 4 , wherein the set of results comprises a detected depth for each object in the projected 3D scene, and wherein the set of expected results comprises a known depth for each object in the 3D scene. |
| Claim: | 6. The verification system of claim 1 , wherein the object detection and classification routine is verified, modified, or disabled based on whether the performance of the object detection and classification routine satisfies a threshold metric. |
| Claim: | 7. The verification system of claim 1 , wherein the projection system comprises a standalone projector that is configured to be temporarily connected to the vehicle. |
| Claim: | 8. The verification system of claim 1 , wherein the projection system comprises a projector that is integrated into the vehicle. |
| Claim: | 9. The verification system of claim 8 , wherein the projector is integrated into a headlight or windshield of the vehicle. |
| Claim: | 10. The verification system of claim 1 , wherein the wall surface is a garage wall or a vehicle service center wall. |
| Claim: | 11. An object detection and classification verification method for a vehicle, the method comprising: providing a projection system configured to project visible light onto a wall surface in front of the vehicle, wherein the wall surface is substantially perpendicular to a ground surface beneath the vehicle; and verifying, by a controller of the vehicle in communication with the projection system, a performance of an object detection and classification routine by: commanding, by the controller, the projection system to project one of a plurality of three-dimensional (3D) scenes onto the wall surface in front of the vehicle, wherein each of the plurality of 3D scenes is pre-captured at a known distance and comprises at least one known object; performing, by the controller, the object detection and classification routine on the projected 3D scene to generate a set of results; comparing, by the controller, the set of results to a set of expected results associated with the projected 3D scene; and based on the comparing, determining, by the controller, whether the performance of the object detection and classification routine satisfies a predetermined threshold metric. |
| Claim: | 12. The method of claim 11 , wherein the set of results comprises a detected location and detected class for each object in the projected 3D scene, and wherein the set of expected results comprises a known location and known class for each object in the 3D scene. |
| Claim: | 13. The method of claim 12 , wherein each of the plurality of 3D scenes further comprises a two-dimensional (2D) box having a size corresponding to the known distance. |
| Claim: | 14. The method of claim 13 , further comprising at least one of: outputting, by the controller, instructions to move the vehicle towards or away from the wall surface until the size of the 2D box in the projected 3D scene matches the size corresponding to the known distance; and outputting, by the controller, instructions to move the vehicle left or right relative to the wall surface until there is no projection distortion of the 2D box in the projected 3D scene. |
| Claim: | 15. The method of claim 14 , wherein the set of results comprises a detected depth for each object in the projected 3D scene, and wherein the set of expected results comprises a known depth for each object in the 3D scene. |
| Claim: | 16. The method of claim 11 , further comprising verifying, modifying, or disabling the object detection and classification routine based on whether the performance of the object detection and classification routine satisfies a threshold metric. |
| Claim: | 17. The method of claim 11 , wherein the projection system comprises a standalone projector that is configured to be temporarily connected to the vehicle. |
| Claim: | 18. The method of claim 11 , wherein the projection system comprises a projector that is integrated into the vehicle. |
| Claim: | 19. The method of claim 18 , wherein the projector is integrated into a headlight or windshield of the vehicle. |
| Claim: | 20. The method of claim 11 , wherein the wall surface is a garage wall or a vehicle service center wall. |
| Patent References Cited: | 11117570 September 2021 Broggi 11334762 May 2022 Wrenninge 11373411 June 2022 Goh 20080129544 June 2008 Augst 20100289632 November 2010 Seder 20100292886 November 2010 Szczerba 20170219706 August 2017 Baldischweiler 20180029641 February 2018 Solar 20180079463 March 2018 Pearce 20190079525 March 2019 Towal 20190317519 October 2019 Chen 20200026282 January 2020 Choe 20220091414 March 2022 Cappuccilli |
| Other References: | Ramanagopal, Manikandasriram Srinivasan, et al., “Failing to Learn: Autonomously Identifying Perception Failures for Self-Driving Cars”, IEEE Robotics and Automation Letters, Preprint Version, 8 pages, Jul. 2018. cited by applicant |
| Primary Examiner: | Shin, Soo |
| Attorney, Agent or Firm: | Klobucar, Jeremy J. |
| رقم الانضمام: | edspgr.11972612 |
| قاعدة البيانات: | USPTO Patent Grants |
| الوصف غير متاح. |