Vehicle Object Tracking With Visibility Maps Under Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle tracking systems struggle with maintaining a historical understanding of their environment, accurately handling uncertainty, and making robust decisions under conditions of occlusion and sparse information, often leading to incorrect classifications and poor decision-making.
Innovation Solution
A system and method that utilizes multiple representations of the vehicle's environment, incorporating a visibility map to assess detection probability and track objects with a probability of existence, allowing for adaptive tracking and decision-making that considers historical and current data, using diverse sensing modalities and probabilistic methods to enhance accuracy and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional tracking systems are used, then the system complexity is low, but the reliability of tracking under occlusion and sparse information deteriorates
Solution Approach 1:
The tracking system is segmented into multiple independent components: visibility map generation module, probability of detection calculation module, probabilistic data association module, and track management module. Each component handles a specific aspect of the tracking problem, allowing the system to maintain high reliability through specialized processing while managing complexity through modular architecture.
Solution Approach 2:
A visibility map is introduced as an intermediary data structure that mediates between sensor measurements and track updates. This visibility map encodes occlusion information and detection probabilities, serving as a bridge that allows the tracking algorithm to account for environmental factors without directly complexifying the core tracking logic.
2Measurement precision
If multiple representations and probabilistic methods are used, then the accuracy of tracking deteriorates under occlusion, but the loss of information increases
Solution Approach 1:
The system performs preliminary actions by generating visibility maps and calculating probability of detection values before actual track updates occur. This advance preparation of environmental context information ensures that when occlusion occurs, the tracker has pre-computed probability data to guide its decisions, maintaining accuracy without requiring additional information during the tracking process.
Solution Approach 2:
The probabilistic data association framework implements feedback mechanisms where track probabilities are continuously updated based on visibility map information and measurement likelihoods. This feedback loop allows the system to maintain accurate tracking by constantly adjusting track existence probabilities based on current environmental conditions and measurement quality, preventing information loss through adaptive decision-making.
3Reliability
If visibility maps and probability of detection are incorporated, then the robustness of decision-making improves, but the computational power required increases
Solution Approach 1:
The visibility map computation and probability of detection calculations are performed locally for each track and each time step rather than globally for the entire scene. This local quality approach allows the system to maintain robust decision-making for individual tracks while reducing overall computational power requirements by avoiding redundant calculations across the entire environment.
Solution Approach 2:
The system changes parameters dynamically by adjusting the level of detail in visibility map computation based on track priority, occlusion severity, and computational resources available. High-priority tracks or those under significant occlusion receive more computationally intensive processing, while less critical tracks use simplified models, optimizing the balance between decision-making robustness and computational power consumption.
Data Source
AI summary
A method can include: determining a set of measurements; determining a visibility representation; determining a set of observations; generating a set of hypotheses; determining a set of tracks; and/or any other suitable elements. Additionally or alternatively, the method can optionally include planning a trajectory for the vehicle and/or any other suitable elements. The method functions to track objects and the uncertainty of existence thereof in the vehicle's environment over time.