Lidar Self-Reflection Filtering Using a 3D Vehicle Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Self-driving vehicles face challenges in autonomous driving mode due to self-reflections of sensor signals, which can lead to false detections of phantom items or altered object locations, caused by reflections off the vehicle itself, especially when the first return on reflective surfaces is missing, such as mirrors.
Innovation Solution
A method involving processors to determine vehicle and sensor pose information, evaluate return signals against a 3D model of the vehicle to identify self-return points, and perform operations such as discarding or flagging these points to prevent false data interpretation, using a static or dynamic 3D model depending on the vehicle type and employing a buffer around vehicle surfaces to account for range precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system checks whether a lidar point is the second or third return of its beam to filter reflections, then false detections from external reflections are reduced, but the system fails when the first return is missing (e.g., on mirror-like surfaces) and cannot distinguish self-reflections effectively
Solution Approach 1:
The patent segments the detection space by creating a 3D model of the vehicle and dividing the environment into regions that are geometrically possible to detect directly versus regions that would require reflection off the vehicle. This segmentation allows the system to identify self-reflections by determining whether detected points lie in reflection zones rather than relying on return order, which fails on mirror-like surfaces.
Solution Approach 2:
The patent introduces a 3D vehicle model as an intermediary representation between the sensor data and the reflection detection logic. This digital model serves as a mediator that enables geometric reasoning about potential reflection paths without requiring direct measurement of reflection characteristics, thus working reliably across all surface types including mirrors.
2Measurement precision
If the system maintains an image of the lidar field of view to identify reflection points, then external reflections can be filtered, but points near the edge of the FOV create ambiguity and self-reflections from the vehicle body cannot be distinguished
Solution Approach 1:
The patent transitions from a 2D FOV image representation to a 3D spatial model of the vehicle. By elevating the detection logic to three dimensions, the system can perform ray-casting and geometric intersection tests that unambiguously identify self-reflections regardless of their position in the FOV, eliminating the edge-case ambiguity present in 2D approaches.
3Ease of operation
If the system uses a static 3D model of the vehicle for self-return evaluation, then processing is simplified with lookup tables, but the system cannot adapt to articulating vehicles with dynamic configurations
Solution Approach 1:
The patent implements a dynamic 3D vehicle model that can adapt its geometry based on the actual vehicle configuration. For articulating vehicles, the model updates its structure to reflect the current relative positions of vehicle segments, enabling accurate self-reflection detection across static and dynamic configurations while maintaining the computational efficiency of model-based approaches.
4Measurement precision
If the system evaluates every received point against the 3D vehicle model to identify self-returns, then self-reflection detection accuracy is improved, but computational load and processing time increase
Solution Approach 1:
The patent applies local quality optimization by using the 3D vehicle model to pre-identify and flag only those detection points that lie in geometric regions consistent with self-reflections. Rather than uniformly processing all points, the system focuses computational resources on evaluating points in high-probability self-reflection zones, maintaining high detection accuracy while reducing overall processing load.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively mitigates self-reflection issues by accurately distinguishing self-return points from actual environmental data, reducing the likelihood of false object detection and ensuring accurate navigation and object recognition in autonomous driving scenarios.
Implementation Method 1
signals from objects in the external environment are reflected off of a portion of the self-driving vehicle
Implementation Method 2
lidar returns
Data Source
AI summary
The technology relates to handling of self-reflections of sensor signals off of a portion of a vehicle that is operating in an autonomous driving mode. Vehicle pose information and sensor pose information are determined at a given point in time while operating in the autonomous driving mode. Return signals from one or more scans of the environment are received from onboard sensors such as lidar sensors. The system evaluates, based on the vehicle and sensor pose information, whether a segment between a given one of the one or more sensors and a received point from a selected one of the return signals crosses any surface of a 3D model of the vehicle. The received point is identified as a self-return point. In response to identifying the received point as a self-return point, the vehicle is able to perform a driving operation in the autonomous driving mode.