Multi-Sensor Obstacle Detection for High-Speed Stationary Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Self-driving vehicles at Level 2 automation struggle to recognize stationary or slowly moving obstacles due to limitations in sensor technology, leading to potential accidents.
Innovation Solution
Employing a combination of a depth image capture unit, laser image capture unit, and optical image capture unit to merge images using a fusion algorithm, enabling the detection of stationary objects through image optical flow methods, and generating an alarm or controlling brakes when necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If Level 2 self-driving vehicles use forward-looking cameras and millimeter-wave radars to sense the environment, then the vehicle can autonomously complete lane changes, ramp entering, and ramp exiting, but the vehicle cannot recognize stationary or slowly moving obstacles, leading to potential accidents
Solution Approach 1:
The patent combines multiple sensor types (cameras, millimeter-wave radars, and additional sensors) to create a comprehensive sensing system. By merging the data from these different sensors, the system achieves both autonomous driving capability and improved stationary obstacle detection, resolving the contradiction between automation level and reliability
Solution Approach 2:
The sensing system is designed to perform multiple functions: detecting moving vehicles, identifying stationary obstacles, and enabling autonomous navigation. This multi-functional approach allows the same sensor suite to support both high-level automation and reliable obstacle recognition across different scenarios
2Ease of operation
If the vehicle uses existing sensor technology and algorithms, then the system can handle common driving scenarios, but it fails to detect stationary objects or objects with low relative speed
Solution Approach 1:
The system changes the detection parameters by introducing new sensor types and adjusting algorithmic thresholds to specifically target stationary and slow-moving obstacles. This allows the system to maintain simplicity for common scenarios while enhancing detection capability for challenging cases through parameter optimization
3Measurement precision
If the vehicle relies on training data sets for object recognition, then the system can identify common targets, but it cannot recognize new irregular targets or objects not in the training data
Solution Approach 1:
The system incorporates feedback mechanisms where detection results from multiple sensors are continuously evaluated and used to refine the recognition algorithms. This feedback loop enables the system to adapt to new target types and improve identification accuracy over time, balancing precision with versatility
Data Source
AI summary
The present invention provides a method for detecting stationary objects by vehicles moving at high speed. The method is applied to a vehicle, which includes an on-board computer, a depth image capture unit, a laser image capture unit, and an optical image capture unit. When the vehicle moves at a first speed, the depth image capture unit, the laser image capture unit, and the optical image capture unit captures a depth image, a laser image, and an optical image. The on-board computer merges the depth image, the laser image, and the optical image according to give a merged image. Then on-board computer analyzes and judges the stationary object in the merged image according to an image optical flow method and generating stationary object information. The on-board computer displays alarm messages and the stationary object on the display unit of the vehicle according to the stationary object information.