Learning to Grasp from a Single Demonstration
This addresses the need for simpler and faster robotic grasping methods, though it appears incremental as it builds on existing learning-from-demonstration techniques.
The paper tackles the problem of reducing data requirements for robotic grasping by proposing a learning-from-demonstration approach that detects objects to grasp from a single demonstration using GraspNet, achieving training with only hundreds of iterations on a Franka Panda cobot.
Learning-based approaches for robotic grasping using visual sensors typically require collecting a large size dataset, either manually labeled or by many trial and errors of a robotic manipulator in the real or simulated world. We propose a simpler learning-from-demonstration approach that is able to detect the object to grasp from merely a single demonstration using a convolutional neural network we call GraspNet. In order to increase robustness and decrease the training time even further, we leverage data from previous demonstrations to quickly fine-tune a GrapNet for each new demonstration. We present some preliminary results on a grasping experiment with the Franka Panda cobot for which we can train a GraspNet with only hundreds of train iterations.