Autonomous Target Following with Predictive Sensor Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing target following methods in robotics lack accuracy and reliability, especially in crowded and cluttered environments, and require significant computational resources or specialized devices, failing to effectively track targets that make frequent turns or move out of the sensors' field of view.
Innovation Solution
An autonomous target following method that uses environmental measurements from sensors to cluster data into segments, extract features, filter for target location, predict future locations, and adjust based on historical data, employing machine learning and Kalman filters to maintain target tracking in complex scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If UWB technology is used to determine target location, then measurement precision is improved, but device complexity increases due to requiring special peripheral devices
Solution Approach 1:
The patent extracts the localization function from specialized peripheral devices and implements it using standard sensors already present on the robot. The sensor data processing pipeline extracts target location information from environmental measurements without requiring additional UWB hardware, thereby removing the complexity of special peripheral devices while maintaining localization capability.
Solution Approach 2:
The patent makes the robot's existing sensors serve multiple functions: they are used for both general environmental perception and specifically for target localization and tracking. This multi-functionality eliminates the need for dedicated UWB peripheral devices, reducing device complexity while maintaining the ability to achieve accurate target location determination.
2Reliability
If computer vision and machine learning techniques are used for target detection, then target recognition capability is improved, but use of energy increases due to huge computational resources required
Solution Approach 1:
The patent segments the target detection process into distinct stages: environmental measurement collection, data clustering into segments, feature extraction from segments, and target identification based on features. This segmentation allows for more efficient processing at each stage, reducing overall computational resource consumption while maintaining reliable target detection accuracy.
Solution Approach 2:
The patent applies partial action by processing only the most relevant features from sensor data rather than performing exhaustive computer vision analysis on entire datasets. By focusing on key features extracted from clustered segments, the system achieves reliable target detection with reduced computational energy consumption compared to full machine learning approaches.
3Device complexity
If traditional localization methods are used, then device complexity is reduced, but measurement precision of target location deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-processing sensor measurements through clustering into segments and extracting features before target location determination. This preparatory processing of data enhances the precision of subsequent location calculations, allowing the system to achieve high measurement precision using only standard sensors without adding device complexity.
Solution Approach 2:
The patent implements feedback by continuously processing sensor measurements, comparing extracted features against known target characteristics, and refining target location estimates based on this comparison. This feedback loop enables the system to achieve high measurement precision for target location using simple standard sensors, compensating for any individual sensor limitations through iterative refinement.
Data Source
AI summary
Systems and methods are provided for autonomous target following. An exemplary method may comprise: receiving environmental measurements collected by one or more sensors equipped on a robotic device; clustering the environmental measurements into a plurality of segments; and determining a location of a target based on the plurality of segments.