Ground Profile Estimation With Reflection-Aware Sensor Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately differentiating between ground and objects due to inaccurate sensor data, particularly from reflective materials like glass and metal, which can lead to unsafe navigation scenarios.
Innovation Solution
A trajectory verification system that includes a secondary perception component to classify sensor data as ground or object, using techniques such as M-estimator and machine-learning models, and corrects the ground profile by excluding sensor data points that contribute to inaccuracies, thereby validating or rejecting vehicle trajectories to prevent collisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If sensor data from reflective materials (glass, metal) is used for ground detection, then more sensor data is available for processing, but measurement precision deteriorates due to multi-path reflections causing aberrations
Solution Approach 1:
The patent extracts and removes inaccurate sensor data points that cause multi-path reflections from the dataset. By identifying and excluding these problematic reflections, the system maintains measurement precision while still utilizing the majority of valid sensor data for ground detection and object differentiation.
Solution Approach 2:
The patent introduces an intermediary processing layer that analyzes sensor data for accuracy before ground profile estimation. This intermediary step identifies and filters out reflective artifacts, allowing the system to mediate between raw sensor data and final ground detection results, thereby maintaining precision without losing valuable data.
2Speed
If all sensor data points are used for ground profile estimation, then processing speed increases, but reliability decreases due to inclusion of inaccurate data from reflections
Solution Approach 1:
The patent performs preliminary filtering of sensor data to identify and exclude inaccurate points before conducting ground profile estimation. This preliminary action ensures that only reliable data points are used in the main processing pipeline, maintaining both speed and reliability by preventing error propagation from the outset.
Solution Approach 2:
The patent implements a feedback mechanism where the system evaluates sensor data quality and adjusts the ground profile estimation process accordingly. By continuously monitoring data accuracy and providing feedback to the processing algorithm, the system maintains reliability without sacrificing processing speed through iterative refinement.
3Measurement precision
If complex machine-learning models are used to classify sensor data, then measurement precision improves for object differentiation, but device complexity increases
Solution Approach 1:
The patent segments the perception system into distinct functional components: data filtering, ground profile estimation, and object classification. By dividing the complex machine-learning task into smaller, specialized modules, the system achieves high measurement precision while managing device complexity through modular architecture and targeted model deployment.
Data Source
AI summary
Techniques for estimating a ground profile for an environment associated with a vehicle are described herein. Sensor data associated with a vehicle may be used to determine sensor data points associated with a path of the vehicle. Using a first metric and a second metric, such as a maximum slope and minimum slope based on the position of the vehicle and/or a previously-determined ground point, a region (e.g., search window) may be determined, and based on the region, a subset of sensor data points included in the region. The subset of sensor data points may be used to determine a ground point. In some instances, the ground point may be determined based on an average height of the sensor data points. A ground profile may be determined that at least partially includes the ground point.