Agent Prediction Scheduling Under Autonomous Vehicle Compute Limits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in generating timely and accurate prediction data for agents in their vicinity due to limited computing resources, leading to sub-optimal scheduling and potential delays in planning decisions, especially in busy environments.
Innovation Solution
A system that uses complexity and importance scoring models to dynamically allocate prediction data generation tasks across available computing resources, prioritizing agents with the greatest impact on planning decisions, thereby optimizing resource utilization and ensuring prompt data generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If prediction data is generated for all agents in the vicinity, then completeness of prediction data is improved, but computing resource consumption increases beyond available capacity
Solution Approach 1:
The patent segments the set of all agents into high-priority and low-priority subsets based on importance scoring. High-priority agents (those with greater impact on planning decisions) receive full prediction data generation, while low-priority agents receive reduced or deferred processing. This segmentation allows the system to allocate limited computing resources effectively across multiple agents without overwhelming the system capacity.
Solution Approach 2:
The patent applies local quality by providing different levels of prediction data processing to different agents based on their individual importance scores. Rather than uniform processing, each agent receives prediction data generation tailored to its significance for planning decisions. High-importance agents receive comprehensive analysis while low-importance agents receive minimal or deferred processing, optimizing overall resource utilization.
2Productivity
If prediction data generation is performed for multiple agents concurrently, then productivity is improved, but data accuracy deteriorates due to resource constraints
Solution Approach 1:
The patent implements dynamic task scheduling that adjusts the level of prediction data generation for each agent based on real-time importance scoring and system load conditions. The scheduler dynamically prioritizes high-importance agents when resources are available, ensuring accurate prediction data, while automatically reducing or deferring low-importance agent processing when computing resources are constrained. This dynamic adjustment maintains accuracy for critical agents while maximizing overall productivity.
Solution Approach 2:
The patent changes the processing parameters (completeness and depth of prediction data generation) based on agent importance scores and system state. For high-importance agents, full prediction models are executed with high accuracy parameters. For low-importance agents, simplified models or reduced-parameter processing is applied. This parameter adjustment allows concurrent processing of multiple agents while maintaining necessary accuracy levels for safety-critical predictions.
3Quantity of substance
If computing resources are allocated to generate prediction data for low-priority agents, then completeness of data coverage is improved, but time for generating data for high-priority agents increases
Solution Approach 1:
The patent performs preliminary action by calculating importance scores for all agents before allocating computing resources for prediction data generation. Based on these pre-calculated importance scores, the system pre-determines the processing priority and resource allocation for each agent. This preliminary classification ensures that high-priority agents are always processed first with full resources, preventing time delays, while low-priority agents are scheduled for processing only when resources become available.
Solution Approach 2:
The patent maintains continuity of useful action by implementing a continuous scheduling mechanism that constantly monitors system load and agent importance. Rather than batch processing all agents equally, the system continuously allocates computing resources to the most important pending prediction tasks. This ensures uninterrupted, timely generation of prediction data for high-priority agents while progressively processing low-priority agents as resources permit, maximizing both time efficiency and data coverage.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining a task schedule for generating prediction data for different agents. In one aspect, a method comprises receiving data that characterizes an environment in a vicinity of a vehicle at a current time step, the environment comprising a plurality of agents; receiving data that identifies high-priority agents for which respective data characterizing the agents must be generated at the current time step; identifying available computing resources at the current time step; processing the data that characterizes the environment using a complexity scoring model to determine a respective complexity score for each of the high-priority agents; and determining a schedule for the current time step that allocates the generation of the data characterizing the high-priority agents across the available computing resources based on the complexity scores.