Lidar Object Pose Correction for Motion-Distorted Scans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle sensor systems struggle to accurately determine the pose of moving objects due to displacement during scans, leading to inaccuracies in geometric container formation and increased processing resources, which degrades the ability to classify and label objects effectively.
Innovation Solution
A vehicle computer utilizes velocity correction techniques, including machine learning, to iteratively adjust amodal representations of moving objects, forming geometric containers with straightened boundaries by compensating for object motion, thereby improving classification and reducing processing resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If velocity correction is applied to improve object pose determination accuracy, then measurement precision improves, but device complexity increases
Solution Approach 1:
The system performs velocity correction as a preliminary step before final pose determination. By calculating velocity compensation based on object displacement between scans and applying it to adjust point cloud data beforehand, the system improves measurement precision while managing complexity through structured preprocessing
Solution Approach 2:
The system introduces velocity compensation as an intermediary calculation that bridges raw sensor data and final pose determination. This intermediary step calculates displacement-based corrections and applies them to align point clouds from multiple scans, improving accuracy without requiring complete system redesign
2Measurement precision
If multiple scans are performed to improve object pose determination, then measurement precision improves, but loss of time increases
Solution Approach 1:
The system performs scans continuously and processes data in real-time rather than waiting for complete scan sets. By continuously acquiring point cloud data and applying velocity correction iteratively, the system improves precision through multiple measurements while minimizing time loss through continuous processing
Solution Approach 2:
The system performs preliminary velocity correction calculations using available scan data before final pose determination. This allows the system to utilize multiple scans for improved precision while reducing overall processing time through staged computation
3Reliability
If velocity correction processing is applied to improve object classification, then reliability improves, but use of energy increases
Solution Approach 1:
The system uses the vehicle's existing motion data and sensor outputs to perform velocity correction without requiring additional dedicated sensors or external resources. By leveraging already-collected scan data and vehicle state information, the system improves classification reliability while minimizing additional energy consumption
Solution Approach 2:
The velocity correction processing serves multiple functions: it improves pose determination accuracy, enhances object classification reliability, and works with both lidar and radar sensor types. This multi-functionality justifies the energy investment by providing broad benefits across different sensing modalities and application scenarios
Data Source
AI summary
A computer includes a processor and a memory, the memory stores instructions executable by the processor to generate first and second sets of points from first and second scans obtained from a lidar sensor, to determine a first velocity-compensated position of an object represented by a third set of points at a first validity time that is between respective times of the first and second scans. The instructions can additionally be to receive a parameter from the memory of the computer, in which the parameters are determined from a training process to modify an amodal representation of the object, the modified amodal representation being determined from a difference between a second velocity-compensated position of the object and an unmodified amodal representation of the object. The instructions can additionally be to determine a pose of the object represented by the third set of points based on the parameter.