Synthetic Point Cloud Generation for Occluded Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle systems face challenges in accurately detecting and recognizing objects at distances beyond a predetermined range and when objects are occluded, which can lead to inadequate prediction data for timely and safe navigation decisions.
Innovation Solution
The system processes a temporal sequence of point cloud frames to generate synthetic data points, which are then used to enhance object detection accuracy by propagating object information from other frames, thereby improving the performance of the object detection system without significant computation overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional object detection techniques are used to process point cloud data, then the system can operate with standard computational resources, but detection accuracy deteriorates for objects beyond predetermined range or that are occluded
Solution Approach 1:
The system performs preliminary object detection on earlier point cloud frames before the target frame, generating predicted locations for objects in the target frame. This preliminary action allows the system to proactively identify objects that may be distant or occluded in the current frame, improving detection accuracy by preparing detection data in advance based on temporal sequences.
2Measurement precision
If the system processes more point cloud frames to improve detection of distant objects, then detection accuracy improves, but computational overhead increases
Solution Approach 1:
The system extracts and processes only the necessary temporal sequence of point cloud frames rather than processing all available frames. By selectively taking out and processing a specific temporal window of frames that contain relevant object information, the system improves detection accuracy for distant objects while avoiding the computational overhead of processing the entire frame sequence.
3Reliability
If the system uses synthetic data points from temporal sequences, then detection of occluded objects improves, but data processing complexity increases
Solution Approach 1:
The system creates synthetic data points that copy and propagate object information from earlier point cloud frames to the target frame. By generating these synthetic copies of object data based on temporal sequences and predicted trajectories, the system improves detection reliability for occluded objects without requiring complex processing of the actual occluded objects themselves.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing point cloud data to generate an output. In one aspect, a method includes obtaining a sequence of multiple point cloud frames, wherein the sequence of multiple point cloud frames comprise a target point cloud frame and a plurality of other point cloud frames; processing one or more other point cloud frames to generate one or more respective predicted locations at the target timestamp for each of one or more objects detected in the one or more other point cloud frames; generating, based on the respective predicted locations at the target timestamp for each of one or more objects, a synthetic point cloud frame that is associated with the target timestamp; and processing at least the synthetic point cloud frame to generate one or more outputs that characterize an environment at the target timestamp.