Actor Importance Model for AV Perception Prioritization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in efficiently processing and prioritizing detected objects in complex environments due to limited computing resources, leading to incomplete assessment of actors within a specified time period.
Innovation Solution
An actor importance model is trained to rank detected actors based on their features and the vehicle's intent, using neural network layers to generate embeddings and determine relative importance, allowing for focused processing on higher-priority actors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the AV applies limited processing to all detected objects within a specified time period, then the processing speed is maintained, but the completeness of environmental assessment deteriorates
Solution Approach 1:
The patent segments the set of detected actors into multiple priority groups based on their importance to the autonomous vehicle. The system identifies high-priority actors (such as those close to the vehicle or moving rapidly) and processes them in full detail, while lower-priority actors receive reduced processing. This segmentation allows the system to maintain high processing speed for critical objects while still assessing a larger portion of the environment, thereby resolving the contradiction between processing speed and completeness of environmental assessment.
2Productivity
If the AV focuses resource usage on higher importance actors, then processing efficiency is improved, but the ability to detect lower importance actors deteriorates
Solution Approach 1:
The patent applies local quality by assigning different processing qualities to different actors based on their individual characteristics and importance. High-priority actors receive comprehensive processing with full computational resources, while lower-priority actors receive streamlined processing with reduced resources. This differentiated approach ensures that critical detection tasks are performed with high reliability while maintaining overall processing efficiency across all detected actors.
3Measurement precision
If the AV ranks actors by importance using complex models, then the accuracy of prioritization is improved, but the computational resource consumption increases
Solution Approach 1:
The patent implements partial action by applying complex prioritization models selectively rather than uniformly to all actors. The system uses sophisticated ranking algorithms for high-priority actors where accurate prioritization is critical, while employing simpler, faster models for lower-priority actors. This partial application of complex models maintains high accuracy where needed while significantly reducing overall computational resource consumption, resolving the contradiction between prioritization accuracy and energy usage.
Data Source
AI summary
An actor importance model predicts an actor importance ranking of actors in the environment of an autonomous vehicle (AV). The actors represent detected objects in the environment that may be relevant to perception and planning of the AV. To direct resources of the AV, the actors are ranked in the relative importance to the AV's intent in navigating the environment. The actor importance model may generate embeddings to represent the actors, AV intent, and the overall scene. To train the model, a relative ordering loss may be used that evaluates the relative ordering of the actors with respect to one another, which may be further modified based on a threshold for which further processes are affected by the ranking.