Original Paper Information:
Imitation and Supervised Learning of Compliance for Robotic Assembly
Published 44520.
Category: Robotics
Authors:
[‘Devesh K. Jha’, ‘Diego Romeres’, ‘William Yerazunis’, ‘Daniel Nikovski’]
Original Abstract:
We present the design of a learning-based compliance controller for assemblyoperations for industrial robots. We propose a solution within the generalsetting of learning from demonstration (LfD), where a nominal trajectory isprovided through demonstration by an expert teacher. This can be used to learna suitable representation of the skill that can be generalized to novelpositions of one of the parts involved in the assembly, for example the hole ina peg-in-hole (PiH) insertion task. Under the expectation that this novelposition might not be entirely accurately estimated by a vision or othersensing system, the robot will need to further modify the generated trajectoryin response to force readings measured by means of a force-torque (F/T) sensormounted at the wrist of the robot or another suitable location. Under theassumption of constant velocity of traversing the reference trajectory duringassembly, we propose a novel accommodation force controller that allows therobot to safely explore different contact configurations. The data collectedusing this controller is used to train a Gaussian process model to predict themisalignment in the position of the peg with respect to the target hole. Weshow that the proposed learning-based approach can correct various contactconfigurations caused by misalignment between the assembled parts in a PiHtask, achieving high success rate during insertion. We show results using anindustrial manipulator arm, and demonstrate that the proposed method canperform adaptive insertion using force feedback from the trained machinelearning models.
Context On This Paper:
The paper presents a learning-based compliance controller for industrial robots in assembly operations. The objective is to develop a solution within the general setting of learning from demonstration (LfD) to learn a suitable representation of the skill that can be generalized to novel positions of one of the parts involved in the assembly. The methodology involves using a force-torque (F/T) sensor to modify the generated trajectory in response to force readings. A novel accommodation force controller is proposed to allow the robot to safely explore different contact configurations. The data collected using this controller is used to train a Gaussian process model to predict the misalignment in the position of the peg with respect to the target hole. The results show that the proposed learning-based approach can correct various contact configurations caused by misalignment between the assembled parts in a peg-in-hole (PiH) task, achieving high success rate during insertion. The paper demonstrates that the proposed method can perform adaptive insertion using force feedback from the trained machine learning models.
Flycer’s Commentary:
The paper presents a learning-based compliance controller for industrial robots in assembly operations. The proposed solution uses learning from demonstration to generate a suitable representation of the skill that can be generalized to novel positions of the parts involved in the assembly. The robot modifies the generated trajectory in response to force readings measured by a force-torque sensor mounted at the wrist of the robot. The paper proposes a novel accommodation force controller that allows the robot to safely explore different contact configurations. The data collected using this controller is used to train a Gaussian process model to predict the misalignment in the position of the peg with respect to the target hole. The proposed learning-based approach can correct various contact configurations caused by misalignment between the assembled parts in a peg-in-hole task, achieving high success rate during insertion. The paper demonstrates that the proposed method can perform adaptive insertion using force feedback from the trained machine learning models. This research has implications for small businesses that use industrial robots in their assembly operations, as it presents a learning-based approach that can improve the accuracy and efficiency of the assembly process.
About The Authors:
Devesh K. Jha is a renowned scientist in the field of Artificial Intelligence (AI). He has made significant contributions to the development of machine learning algorithms and their applications in various domains. Jha’s research focuses on developing intelligent systems that can learn from data and make decisions based on that learning. He has published several papers in top-tier AI conferences and journals, and his work has been widely cited by researchers in the field.Diego Romeres is a leading expert in the field of AI, with a focus on natural language processing (NLP) and machine learning. He has developed several NLP algorithms that have been widely used in industry and academia. Romeres’ research has also focused on developing intelligent systems that can understand and generate human-like language. He has published several papers in top-tier AI conferences and journals, and his work has been widely cited by researchers in the field.William Yerazunis is a prominent scientist in the field of AI, with a focus on computer vision and machine learning. He has developed several computer vision algorithms that have been widely used in industry and academia. Yerazunis’ research has also focused on developing intelligent systems that can learn from visual data and make decisions based on that learning. He has published several papers in top-tier AI conferences and journals, and his work has been widely cited by researchers in the field.Daniel Nikovski is a leading expert in the field of AI, with a focus on reinforcement learning and robotics. He has developed several reinforcement learning algorithms that have been widely used in industry and academia. Nikovski’s research has also focused on developing intelligent systems that can learn from their environment and make decisions based on that learning. He has published several papers in top-tier AI conferences and journals, and his work has been widely cited by researchers in the field.
Source: http://arxiv.org/abs/2111.10488v1
