In-Hand Pose Refinement for Real-Time Robotic Pick-and-Place
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current robotic systems for industrial assembly face challenges in accurately tracking the six-degrees-of-freedom (6DoF) pose of objects during movement, requiring human intervention and being computationally intensive, which increases cycle time and poses safety hazards.
Innovation Solution
A pose tracking system that includes a moveable camera system programmed to track the robotic arm, using a method that determines an initial pose, calculates an expected pose based on position and velocity data, and refines the pose using image data from a stereo-pair camera system, employing algorithms like ICP for accurate alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing pose estimation techniques are used to achieve high accuracy, then measurement precision is improved, but computational complexity increases making them unsuitable for real-time tracking
Solution Approach 1:
The system performs preliminary action by predicting the object pose at the current frame based on the previous pose and object velocity, before performing refinement. This prediction step provides a good initial estimate that reduces the computational burden of the subsequent refinement process, enabling real-time tracking while maintaining high accuracy.
Solution Approach 2:
The system creates a simplified copy of the pose estimation process by using a predictive model that replicates the essential behavior of complex pose estimation techniques. Instead of running full computational pose estimation algorithms at every frame, the system uses a lightweight predictive copy that maintains accuracy while reducing computational complexity.
2Reliability
If human intervention is used to monitor assembly processes, then reliability is improved, but productivity decreases due to added cycle time and safety hazards
Solution Approach 1:
The system implements self-service by providing automated pose tracking and verification that operates independently without human intervention. The pose tracking system automatically monitors the robotic arm movements and object placement, providing real-time feedback and ensuring reliability through automated detection rather than human monitoring, thus maintaining productivity.
Solution Approach 2:
The system uses feedback by continuously tracking the object pose during movement and comparing it against expected trajectories. This automated feedback mechanism ensures reliability by detecting deviations or potential failures in real-time, replacing the need for human monitoring while maintaining or improving assembly process reliability and productivity.
3Device complexity
If a stationary camera with limited field of view is used, then device complexity is reduced, but measurement precision deteriorates as it cannot capture the entire range of motion
Solution Approach 1:
The system applies dynamics by making the camera system movable rather than stationary. The camera moves together with the robotic arm to maintain a consistent view of the object throughout its range of motion. This dynamic camera configuration enables continuous accurate pose tracking without requiring a complex stationary multi-camera system, balancing device complexity with measurement precision.
4Measurement precision
If computationally intensive pose estimation algorithms are used, then measurement precision is improved, but productivity decreases due to increased processing time
Solution Approach 1:
The system performs preliminary action by predicting the object pose before refinement, providing a head start that reduces the computational work needed during the actual tracking frame. This predictive preprocessing enables the system to achieve high measurement precision while maintaining real-time tracking capability by minimizing processing time per frame.
Solution Approach 2:
The system applies parameter changes by transitioning from full computational pose estimation to a hybrid approach using prediction and selective refinement. This changes the computational parameters from intensive continuous estimation to lighter predictive modeling with occasional refinement, maintaining accuracy while improving productivity for real-time applications.
Data Source
AI summary
A method of tracking a pose of an object includes determining an initial pose of the object at a first position, receiving position data and velocity data corresponding to movement of the object to a second position by a moving device, determining an expected pose of the object at the second position based on the position and velocity data and the initial pose, receiving second image data corresponding to the object at the second position from a camera, and determining a refined pose of the object at the second position based on the second image data and the expected pose.