Novel vision-based “ViseJoint" for remote sensing

ViseJoint is a virtual, software-based sensor based on computer vision that measures angular positions of joints (angular encoder). Multiple joints and even kinematic chains are measured simply by looking at them with a camera. Sensing and control works remotely and is purely camera-based. Like that, mechanic components do not need to integrate any sensor, nor electronics or wiring.

The system is available as a software package for any standard PC and for many embedded systems. It connects to most standard cameras (USBVision, UVC, Ethernet) and provides angular readings for all observed joints. The user only needs to attach textured markers onto the joints. A development kit is also available to get started right away.

Applications include mechanical systems with one or multiple rotational joints that require angular position control, as long as remote sensing is feasible. This includes automation systems, in which multiple components with rotational axes may be monitored by a single camera. For the control of robot arms, especially for service robotics, camera-based encoders measure chains of multiple joints with a pair of markers. Camera-based encoders can also be used for calibration tasks, e.g. to initialise a relative encoder, or as a second channel in conjunction with existing encoders.

Visevi Robotics GmbH is a startup from Munich, Germany, that focuses on computer vision products for sensing and for grasping on smart robots. The patented technology of camera-based sensing is available for tactile data, force-torque and joint angles and has been developed with Technical University of Munich.

Applications

• Angular sensing of any actuated or non-actuated joint
• Automation systems
• Turntables and similar mechanic components
• Calibration tasks
• Robot arms and service robots

Advantages

• Remote sensing with camera placed at central location
• Simplified hardware: No electronic sensor on the joint, no wiring required
• Multiple virtual encoders with one camera
• Co-use of camera for other vision tasksbo

Specifications