Multi-Camera Agent Trajectory Derivation Without LiDAR
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for collecting prior agent trajectories, such as those using LiDAR-based sensor systems, are expensive and limited in scalability due to the high cost of equipment and restricted geographic availability, making it impractical to collect accurate and large-scale data for applications like autonomous driving and transportation matching platforms.
Innovation Solution
The use of monocular cameras and stereo camera systems, integrated with telematics sensors, to capture and process image data for deriving agent trajectories, which involves identifying tracking points, estimating depth, and applying motion models to correct for occlusions and inaccuracies, allowing for more affordable and widespread data collection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR-based sensor systems are used to collect agent trajectories, then measurement precision is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent uses monocular camera images as a simplified copy or alternative representation of the complex LiDAR point cloud data. Instead of directly processing expensive LiDAR data, the system captures 2D image data that can be processed to derive trajectory information, effectively creating a cheaper substitute that maintains sufficient accuracy for the application
Solution Approach 2:
The patent replaces expensive, complex LiDAR sensors with inexpensive monocular cameras that are already widely available in consumer vehicles. This substitution uses cheap, readily available hardware to achieve the same functional goal of collecting trajectory data at scale, making the system economically viable for large-scale deployment
2Measurement precision
If LiDAR-based sensor systems are deployed, then trajectory data accuracy is improved, but scalability deteriorates due to high cost and limited geographic availability
Solution Approach 1:
The patent leverages the universality of monocular cameras, which are already present in most modern vehicles for basic functions like dashcams or driver assistance. By repurposing this existing, universal component for trajectory data collection, the system achieves scalability without requiring specialized equipment deployment across different geographic regions
Solution Approach 2:
By using inexpensive monocular cameras instead of expensive LiDAR systems, the patent enables widespread deployment across many vehicles in diverse geographic locations. The low cost of the camera system removes the economic barrier that limits LiDAR scalability, allowing for large-scale data collection from diverse real-world scenarios
3Device complexity
If monocular cameras are used instead of LiDAR, then device cost decreases and scalability improves, but measurement precision and depth estimation accuracy worsen
Solution Approach 1:
The patent addresses the monocular camera's inherent lack of depth information by introducing temporal dimensionality. Instead of relying on single-frame depth cues, the system tracks agents across multiple frames and uses motion models to infer 3D trajectory from 2D position sequences, effectively compensating for the missing depth dimension through time-based analysis
Solution Approach 2:
The patent introduces motion models as an intermediary computational layer that bridges the gap between 2D monocular image data and 3D trajectory estimation. These models use physical constraints and motion patterns to generate plausible 3D trajectories from 2D observations, acting as a mediator that translates limited visual information into accurate trajectory data
4Loss of information
If tracking is performed on partially occluded agents, then trajectory completeness is improved, but measurement precision deteriorates due to occlusion handling challenges
Solution Approach 1:
The patent applies motion models in advance to predict agent positions during occlusion periods. By using the agent's historical motion patterns and physical constraints, the system can infer missing trajectory points before they are lost to occlusion, maintaining trajectory completeness without relying on potentially inaccurate visual data during occluded periods
Data Source
AI summary
Examples disclosed herein may involve a computing system that is operable to (i) receive a first sequence of images captured by a monocular camera associated with a vehicle during a given period of operation and a second sequence of image pairs captured by a stereo camera associated with the vehicle during the given period of operation, (ii) derive, from the first sequence of images captured by the monocular camera, a first track for a given agent that comprises a first sequence of position information for the given agent, (iii) derive, from the second sequence of image pairs captured by the stereo camera, a second track for the given agent that comprises a second sequence of position information for the given agent, and (iv) determine a trajectory for the given agent based on the first and second tracks for the given agent.