Vision-Based Force-Position Fusion for Dexterous Hand
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional solutions for measuring force and position in dexterous hands are costly, bulky, and complex, requiring multiple sensors and increased power consumption, which restricts the size and stability of the control system.
Innovation Solution
A smart manipulator with vision-based force and position fusion measurement, utilizing a truncated cone structure with depth cameras and drive measurement units, allows for simultaneous measurement of torque and angle information across all joints using a binocular camera, eliminating the need for traditional sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional angle encoders and torque sensors are installed on each joint of the dexterous hand, then force and position measurement precision is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent replaces traditional mechanical sensors (angle encoders and torque sensors) with a vision-based measurement system. Depth cameras capture images of marker points on the dexterous hand, and computer vision algorithms calculate joint angles and forces from these images, substituting mechanical sensing with optical sensing and computational processing.
Solution Approach 2:
The patent uses visual copying by capturing images of marker points attached to the dexterous hand joints. Instead of directly measuring physical quantities with sensors, the system creates visual copies (images) of the joint positions and uses these copies to infer angle and force information through image processing and geometric calculations.
2Measurement precision
If multiple sensors are installed on each joint for force and position measurement, then measurement accuracy is improved, but the overall size and weight of the mechanism increase
Solution Approach 1:
The patent replaces heavy mechanical sensors with lightweight depth cameras and computational algorithms. The cameras are positioned externally to capture images of the dexterous hand, eliminating the need to mount heavy torque sensors and angle encoders on each joint, thereby significantly reducing the weight of the moving dexterous hand mechanism.
3Reliability
If traditional force and position sensing links are used in the manipulator system, then measurement reliability is improved, but the overall mechanism size becomes bulky
Solution Approach 1:
The patent replaces bulky mechanical sensing links with a compact vision-based system. Depth cameras capture three-dimensional information optically, and computational algorithms process these images to extract joint angle and force data, eliminating the need for extensive mechanical sensor installations and reducing the overall volume of the manipulator system.
Solution Approach 2:
The patent transitions from mechanical measurement in physical space to optical measurement in image space. By capturing three-dimensional joint position information through depth camera images and calculating forces through computational geometry, the system achieves reliable measurement without requiring additional physical space for mechanical sensing components.
4Measurement precision
If multiple sensors and wiring are installed on each joint, then measurement capability is improved, but power consumption increases
Solution Approach 1:
The patent replaces power-hungry electronic sensors with optical depth cameras and computational processing. The cameras capture images passively using ambient or active illumination, and the computational algorithms run on external processing units, eliminating the need for power-intensive sensors on each joint of the dexterous hand.
Data Source
AI summary
A smart manipulator based on vision-based force and position fusion measurement, which solves overall bloated structure in the existing manipulator system, includes a thumb, an index finger, a middle finger, a ring finger, a little finger, a palm, two depth cameras and a plurality of drive measurement units, wherein the drive measurement units are evenly distributed to form a truncated cone structure, the two depth cameras are symmetrically arranged on a rear end of the drive measurement units while each depth camera is respectively arranged along a generatrix direction of the truncated cone structure. The palm is arranged at a front end of the drive measurement units, and the five fingers are arranged at a front end of the palm, each of the thumb, the index finger, the middle finger, the ring finger and the little finger are connected to two or more the drive measurement unit respectively.