Spatial Features-Based Slip Detection in Neuromorphic Vision Tactile Sensors

Rapidly emerging neuromorphic technologies are revolutionizing intelligent robotics and human-machine interactions in numerous industrial and healthcare applications. In robotic manipulation, detecting incipient slip is crucial for maintaining grip stability and object safety. This work proposes a novel framework that employs a bioinspired neuromorphic vision-based tactile sensor to detect incipient slip by analyzing spatial features. The proposed spatial features enable discrimination of incipient slip from other robotic manipulation actions, such as pressing and proximity, under minimal contact conditions on rigid, flat, and smooth surfaces. A Random Forest classifier is trained to perform classification based on the extracted spatial features. Experimental validation on unseen datasets demonstrates an overall accuracy of 94.19%, highlighting the framework's potential to enhance the precision of robotic manipulation applications.