Multi-Stage Object Heading Estimation for Occluded Vehicle Targets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately estimating the heading of nearby objects due to the limitations of individual sensing techniques, which can be less effective in different contexts, such as occluded or slow-moving objects, leading to reduced reliability and accuracy in navigation decisions.
Innovation Solution
A two-stage object heading estimation system that combines preliminary heading estimations from various sensors and techniques, including point-cloud, road-based, and motion-based methods, using neural networks and decision trees to generate refined heading estimates, improving accuracy through temporal filtering and machine-learning models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single sensing technique is used for heading estimation, then the system complexity is low, but the accuracy and reliability are reduced especially for occluded or slow-moving objects
Solution Approach 1:
The patent combines multiple sensing techniques (LIDAR, RADAR, camera-based methods) into a unified heading estimation system. Each sensor type processes data independently to generate preliminary heading estimates, which are then integrated through a neural network to produce a refined final estimate. This merging approach resolves the contradiction by achieving high accuracy through multi-sensor fusion while managing complexity through modular architecture and automated neural network processing.
2Reliability
If multiple preliminary heading estimation subsystems are used, then the reliability is improved, but the device complexity increases
Solution Approach 1:
The system employs a neural network that automatically processes multiple preliminary heading estimates from different sensing techniques and self-determines the optimal refined heading estimate. The neural network autonomously weights and integrates the inputs without requiring complex manual configuration or intervention, thereby improving reliability through multiple estimation paths while keeping the system manageable through self-service automation.
3Measurement precision
If temporal filtering and machine-learning models are applied, then the measurement precision is enhanced, but the processing time and computational complexity increase
Solution Approach 1:
The system performs preliminary heading estimations using multiple independent sensing techniques before the final refinement stage. Each sensing technique (LIDAR, RADAR, camera) independently generates preliminary estimates in parallel, preparing the data in advance for the neural network's refinement process. This preliminary action approach enhances precision through thorough multi-stage processing while managing time loss through parallel computation of preliminary estimates.
Data Source
AI summary
Systems, methods, devices, and techniques for generating object-heading estimations. In one example, methods include actions of receiving sensor data representing measurements of an object that was detected within a proximity of a vehicle; processing the sensor data with one or more preliminary heading estimation subsystems to respectively generate one or more preliminary heading estimations for the object; processing two or more inputs with a second heading estimation subsystem to generate a refined heading estimation for the object, the two or more inputs including the one or more preliminary heading estimations for the object; and providing the refined heading estimation for the object to an external processing system.