Vehicle Object Detection Grid Cell Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting and tracking objects in a vehicle's surroundings using sensor data are not precise and robust, particularly for dynamic objects, as they struggle with accurate object identification and speed prediction.
Innovation Solution
A method utilizing a processing unit to analyze sensor data from ambient sensors, forming a grid of cells to ascertain object probabilities and speeds, forming cell clusters based on dynamic cell measures, and expanding clusters geometrically to improve detection and tracking accuracy, while considering previous object states for robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current object detection and tracking methods are used, then the system can identify objects in the surroundings, but the precision and robustness of detection and tracking are insufficient
Solution Approach 1:
The patent divides the surroundings into a grid of cells, where each cell independently stores measurement data including object probability and speed information. This segmentation allows precise local measurements while maintaining overall system robustness through distributed data storage across multiple cells.
Solution Approach 2:
The patent changes the parameter storage approach by maintaining multiple parameters (object probability, speed, occupancy probability) for each cell rather than single-value representations. This enables more precise object characterization and improves detection reliability by providing multiple indicators for object presence and motion.
2Speed
If cell speed information is used for object tracking, then dynamic object detection is improved, but the complexity of data processing increases
Solution Approach 1:
By dividing the environment into discrete cells, the patent simplifies speed processing by associating velocity information with specific spatial locations rather than tracking continuous object trajectories. Each cell independently processes speed data, reducing overall computational complexity.
Solution Approach 2:
The patent implements dynamic cell classification where cells are automatically labeled as static or dynamic based on speed thresholds. This dynamic adaptation allows the system to focus processing resources on moving objects while simplifying handling of stationary elements, improving speed prediction without proportionally increasing complexity.
3Reliability
If occupancy probability is calculated based on previous object states, then tracking robustness is improved, but the time required for detection increases
Solution Approach 1:
The patent pre-calculates and stores occupancy probabilities in each cell based on historical object states and motion models. This preliminary computation allows the system to quickly retrieve preprocessed probability data during real-time detection, improving tracking robustness without incurring computational delays during critical detection phases.
Solution Approach 2:
Each cell independently maintains and updates its own occupancy probability based on local measurement data and predefined transition models. This self-service approach distributes computational burden across multiple independent cell units, reducing centralized processing time while maintaining robust probabilistic tracking.
Data Source
AI summary
A method for detecting and/or tracking an object in surroundings of a vehicle is described. The method ascertains, at a first time, on the basis of sensor data from one or more ambient sensors of the vehicle, measurement data for a multiplicity of cells of a grid of the surroundings of the vehicle, the measurement data for a first cell indicating an object probability. Moreover, the method ascertains an occupancy probability of the first cell being occupied by an object that was already detected at a preceding time. The method further assigns the first cell to the object in dependence on the object probability and in dependence on the occupancy probability.