Vehicle Entity Tracking With Compressed Motion Signatures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving systems face challenges in detecting obstacles, such as potholes, that are not adequately represented in roadmaps and may not be detected by sensors in time for evasive action, especially under conditions of limited visibility, high speed, or driver alertness.
Innovation Solution
A method involving a processing circuitry that generates a signature of the driving environment through dimension expansion and merge operations, allowing for efficient tracking and representation of obstacles, even when they are not well-represented in sensor data, by using a multidimensional representation and reducing power consumption through irrelevant element shutdown.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional sensor-based obstacle detection is used, then the system can detect obstacles in the driving environment, but it fails to detect obstacles like potholes that are not adequately represented in sensor data or roadmaps
Solution Approach 1:
The patent transforms 2D image data into a 3D signature space by extracting multiple features (location, shape, texture, color) and organizing them into hierarchical clusters. This dimensional transformation allows the system to represent obstacles in a more comprehensive feature space, enabling detection of obstacles like potholes that are not distinguishable in raw sensor data.
Solution Approach 2:
The patent introduces signature clusters as an intermediary representation between raw sensor data and obstacle detection decisions. These clusters serve as a mediator that aggregates similar obstacle patterns and enables the system to recognize and detect obstacle types that may not be immediately apparent in the original sensor input.
2Measurement precision
If comprehensive obstacle tracking is performed using multiple images and features, then the system can improve detection accuracy, but the computational complexity and processing time increase
Solution Approach 1:
The patent segments the complex obstacle detection task into distinct processing stages: feature extraction, signature generation, clustering, and matching. By dividing the problem into these manageable segments, the system can process each stage independently with optimized algorithms, reducing overall computational complexity while maintaining high tracking precision.
Solution Approach 2:
The patent performs preliminary actions by pre-computing and storing obstacle signatures and their cluster associations before actual detection occurs. This preprocessing allows the system to quickly match new sensor data against pre-organized signature clusters during real-time operation, reducing computational burden during critical detection phases.
3Loss of information
If the system processes all spanning elements for signature generation, then complete obstacle information is captured, but power consumption increases
Solution Approach 1:
The patent applies local quality by determining the relevancy of each spanning element individually and selectively processing only those elements that contribute significantly to obstacle signature generation. This selective processing approach maintains information completeness for relevant features while eliminating unnecessary computation of irrelevant elements, thereby reducing power consumption.
Solution Approach 2:
The patent implements partial action by processing only the necessary subset of spanning elements required for adequate obstacle detection rather than exhaustively processing all possible elements. This approach achieves sufficient detection accuracy with reduced computational effort and lower energy consumption.
Data Source
AI summary
A method for tracking after an entity, the method may include tracking, by a monitor of a vehicle, a movement of an entity that appears in various images acquired during a tracking period; generating, by a processing circuitry of the vehicle, an entity movement function that represents the movement of the entity during the tracking period; generating, by the processing circuitry of the vehicle, a compressed representation of the entity movement function; and responding to the compressed representation of the entity movement function.