Autonomous Vehicle Object Detection for Precision-Recall Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in safely navigating environments by efficiently detecting and avoiding both common and rare objects, as existing systems require significant processing power and time to analyze objects observed in high precision and high recall detection.
Innovation Solution
The system employs a combination of high precision detection and high recall detection using separate sensors, where high precision detection identifies classifiable objects with accuracy and high recall detection identifies unknown objects with higher sensitivity, and the processor filters out duplicates to generate efficient candidate trajectories for the vehicle.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high precision detection and high recall detection are both performed using separate sensors, then detection accuracy and sensitivity are improved, but processing time and computational load increase
Solution Approach 1:
The patent segments the detection task into two distinct pathways: high precision detection for classifiable objects and high recall detection for unknown objects. This segmentation allows each detection pathway to be optimized independently and processed in parallel, reducing overall processing time while maintaining high detection accuracy for both common and rare objects
Solution Approach 2:
The system performs partial detection actions by applying high precision detection only to objects that can be classified by known classifiers, while using high recall detection for all objects. This selective application of detection methods reduces unnecessary computational overhead while ensuring that critical unknown objects are not missed
2Measurement precision
If high precision detection is performed on all objects, then detection accuracy is improved, but processing speed decreases
Solution Approach 1:
The patent applies different detection qualities to different objects based on their characteristics. High precision detection with known classifiers is applied locally to objects that can be identified, while high recall detection is applied to unknown objects. This local differentiation of detection quality optimizes both accuracy and processing speed by avoiding unnecessary high-precision processing for all objects
3Reliability
If high recall detection is performed on all objects, then detection sensitivity is improved, but false positives increase and processing load increases
Solution Approach 1:
The patent introduces known classifiers as an intermediary between high recall detection and final object identification. High recall detection first identifies potential objects including unknown ones, then known classifiers act as intermediaries to verify and classify detectable objects. This intermediary step reduces false positives from high recall detection while maintaining detection sensitivity for unknown objects
Data Source
AI summary
This disclosure provides systems and methods for controlling a vehicle. The method comprises receiving data from a set of sensors, wherein the data represents objects or obstacles in an environment of the autonomous vehicle; identifying objects or obstacles from the received data; determining multiple sets of attributes of the objects or obstacles, wherein each set of attributes of the objects or obstacles are determined based on data received by an individual sensor; determining a candidate trajectory for the autonomous vehicle based on the multiple sets of attributes of the objects or obstacles; and controlling the autonomous vehicle according to the candidate trajectory.