Search

CN-116206180-B - Perception result processing method and device, storage medium, electronic equipment and vehicle

CN116206180BCN 116206180 BCN116206180 BCN 116206180BCN-116206180-B

Abstract

The invention relates to a perception result processing method, a device, a storage medium, electronic equipment and a vehicle, wherein the method comprises the steps of obtaining a true value perception result corresponding to each frame of environment image, wherein the environment image is an image around the vehicle acquired in the running process of the vehicle; acquiring a first sensing result corresponding to each frame of the environment image determined by a first sensing system and a second sensing result corresponding to each frame of the environment image determined by a second sensing system; and according to the true value sensing results, aligning the first time stamps corresponding to the first sensing results with the second time stamps corresponding to the second sensing results. Therefore, under the condition that the time stamps are matched, the first sensing result and the second sensing result can be compared, and the comparison evaluation of the first sensing system and the second sensing system can be realized.

Inventors

  • ZHANG QIONG

Assignees

  • 小米汽车科技有限公司

Dates

Publication Date
20260505
Application Date
20230222

Claims (10)

  1. 1. A method for processing a perceived result, comprising: Acquiring a true value sensing result corresponding to each frame of environment image, wherein the environment image is an image around the vehicle acquired in the running process of the vehicle; Acquiring a first sensing result corresponding to each frame of the environment image determined by a first sensing system and a second sensing result corresponding to each frame of the environment image determined by a second sensing system; aligning first time stamps corresponding to the first sensing results with second time stamps corresponding to the second sensing results according to the true sensing results; The aligning, according to the true value sensing results, the first time stamps corresponding to the first sensing results with the second time stamps corresponding to the second sensing results includes: determining a truth detection track of the detected target object according to a plurality of truth sensing results; determining a first detection track of the detected target object according to a plurality of first perception results; determining a second detection track of the detected target object according to a plurality of second perception results; And aligning the first time stamps corresponding to the plurality of first perception results with the second time stamps corresponding to the plurality of second perception results according to the true value detection track, the first detection track and the second detection track.
  2. 2. The method of claim 1, wherein the aligning a first timestamp corresponding to the plurality of first perceived results with a second timestamp corresponding to the plurality of second perceived results according to the true value detection trace, the first detection trace, and the second detection trace comprises: according to a first preset duration, determining a first true value detection track to be matched from the true value detection tracks according to the first detection track and the second detection track; according to a second preset duration, determining a matched target perception result from the first true detection track according to the first detection track and the second detection track, wherein the second preset duration is smaller than the first preset duration; And updating the first time stamps corresponding to the plurality of first sensing results and the second time stamps corresponding to the plurality of second sensing results according to the time stamps corresponding to the target sensing results.
  3. 3. The method according to claim 2, wherein updating the first timestamps corresponding to the plurality of first sensing results and the second timestamps corresponding to the plurality of second sensing results according to the timestamps corresponding to the target sensing results comprises: According to the time stamp corresponding to the target sensing result, adjusting a first time stamp corresponding to a first target sensing result matched with the target sensing result in the first detection track and a second time stamp corresponding to a second target sensing result matched with the target sensing result in the second detection track; According to a first timestamp difference value corresponding to the first target sensing result, adjusting a first timestamp corresponding to a first residual sensing result, wherein the first residual sensing result comprises a first sensing result except the first target sensing result in a plurality of first sensing results, the first timestamp difference value is a difference value between an updated first target timestamp and a first target timestamp before updating, and the first target timestamp is a first timestamp corresponding to the first target sensing result; And adjusting a second timestamp corresponding to a second residual sensing result according to a second timestamp difference value corresponding to the second target sensing result, wherein the second residual sensing result comprises a plurality of second sensing results except the second target sensing result, the second timestamp difference value is a difference value between an updated second target timestamp and a second target timestamp before updating, and the second target timestamp is a second timestamp corresponding to the second target sensing result.
  4. 4. A method according to claim 3, characterized in that the method further comprises: determining a second truth detection track from the truth detection tracks according to the position information of the target object in the truth detection tracks; The aligning the first time stamps corresponding to the plurality of first sensing results with the second time stamps corresponding to the plurality of second sensing results according to the true value detection track, the first detection track and the second detection track includes: And aligning the first time stamps corresponding to the plurality of first perception results with the second time stamps corresponding to the plurality of second perception results according to the second true value detection track, the first detection track and the second detection track.
  5. 5. The method according to claim 1, wherein the obtaining the true value sensing result corresponding to each frame of the environmental image comprises: inputting the environment image into a pre-trained environment perception model aiming at the environment image of each frame so as to acquire the truth value perception result output by the environment perception model.
  6. 6. The method according to any one of claims 1-5, further comprising: Determining the accuracy of the first sensing system according to the multiple true value sensing results and the first time stamps after the multiple first sensing results are aligned; and determining the accuracy of the second sensing system according to the plurality of true value sensing results and the second time stamp after the plurality of second sensing results are aligned.
  7. 7. A perception result processing apparatus, comprising: the first acquisition module is configured to acquire a true value sensing result corresponding to each frame of environment image, wherein the environment image is an image around the vehicle acquired in the vehicle driving process; the second acquisition module is configured to acquire a first perception result corresponding to the environment image of each frame determined by the first perception system and a second perception result corresponding to the environment image of each frame determined by the second perception system; an alignment module configured to align, according to the plurality of true value sensing results, first time stamps corresponding to the plurality of first sensing results with second time stamps corresponding to the plurality of second sensing results; The alignment module is further configured to: determining a truth detection track of the detected target object according to a plurality of truth sensing results; determining a first detection track of the detected target object according to a plurality of first perception results; determining a second detection track of the detected target object according to a plurality of second perception results; And aligning the first time stamps corresponding to the plurality of first perception results with the second time stamps corresponding to the plurality of second perception results according to the true value detection track, the first detection track and the second detection track.
  8. 8. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the steps of the method of any of claims 1-6.
  9. 9. An electronic device, comprising: a memory having a computer program stored thereon; A processor for executing the computer program in the memory to implement the steps of the method of any one of claims 1-6.
  10. 10. A vehicle comprising the electronic device of claim 9, or the vehicle and the electronic device of claim 9 are independent of each other and are in communication connection.

Description

Perception result processing method and device, storage medium, electronic equipment and vehicle Technical Field The disclosure relates to the technical field of automatic driving, and in particular relates to a perception result processing method, a device, a storage medium, electronic equipment and a vehicle. Background With the development of artificial intelligence technology, the automatic driving technology of vehicles is also rapidly developing. The sensors on the automatic driving vehicle generally comprise a camera sensor, a laser radar sensor, a solid-state radar sensor and the like, the sensing system is required to output a sensing result according to image data acquired by the sensors, and the automatic driving system controls the automatic driving of the vehicle according to the sensing result. It can be seen that the perception system is critical in autopilot. In the related art, the performance of the sensing system of the vehicle is evaluated through a large number of drive tests or recharging simulation tests, but the mode can only test the performance of the sensing system, and the gap between the sensing system and other sensing systems in the field can not be determined. Therefore, how to realize the comparison test of different sensing systems is a problem to be solved. Disclosure of Invention In order to overcome the problems in the related art, the present disclosure provides a sensing result processing method, a sensing result processing device, a storage medium, an electronic device, and a vehicle. According to a first aspect of an embodiment of the present disclosure, there is provided a sensing result processing method, including: Acquiring a true value sensing result corresponding to each frame of environment image, wherein the environment image is an image around the vehicle acquired in the running process of the vehicle; Acquiring a first sensing result corresponding to each frame of the environment image determined by a first sensing system and a second sensing result corresponding to each frame of the environment image determined by a second sensing system; and according to the true value sensing results, aligning the first time stamps corresponding to the first sensing results with the second time stamps corresponding to the second sensing results. Optionally, the aligning the first time stamps corresponding to the first sensing results with the second time stamps corresponding to the second sensing results according to the true sensing results includes: determining a truth detection track of the detected target object according to a plurality of truth sensing results; determining a first detection track of the detected target object according to a plurality of first perception results; determining a second detection track of the detected target object according to a plurality of second perception results; And aligning the first time stamps corresponding to the plurality of first perception results with the second time stamps corresponding to the plurality of second perception results according to the true value detection track, the first detection track and the second detection track. Optionally, the aligning the first time stamps corresponding to the plurality of first sensing results with the second time stamps corresponding to the plurality of second sensing results according to the true value detection track, the first detection track and the second detection track includes: according to a first preset duration, determining a first true value detection track to be matched from the true value detection tracks according to the first detection track and the second detection track; according to a second preset duration, determining a matched target perception result from the first true detection track according to the first detection track and the second detection track, wherein the second preset duration is smaller than the first preset duration; And updating the first time stamps corresponding to the plurality of first sensing results and the second time stamps corresponding to the plurality of second sensing results according to the time stamps corresponding to the target sensing results. Optionally, the updating the first time stamps corresponding to the plurality of first sensing results and the second time stamps corresponding to the plurality of second sensing results according to the time stamps corresponding to the target sensing results includes: According to the time stamp corresponding to the target sensing result, adjusting a first time stamp corresponding to a first target sensing result matched with the target sensing result in the first detection track and a second time stamp corresponding to a second target sensing result matched with the target sensing result in the second detection track; According to a first timestamp difference value corresponding to the first target sensing result, adjusting a first timestamp corresponding to a first residual sensing