This paper presents a state-of-the-art survey on robotic systems, sensors, actuators and collaborative strategies for Physical Human-Robot Collaboration (pHRC). The paper starts with an overview of some robotic systems with cuttingedge technologies (sensors and actuators) suitable for pHRC operations and the intelligent assist devices employed in pHRC. Sensors being among the essential components to establish communication between a human and a robotic system are surveyed. The sensor supplies the signal needed to drive the robotic actuators. The survey reveals that the design of new generation collaborative robots and other intelligent robotic systems has paved the way for sophisticated learning techniques and control algorithms to be deployed in pHRC. Furthermore, it revealed relevant components needed to be considered for effective pHRC to be accomplished. Finally, a discussion of the major advances made, some research directions, and future challenges are presented.

At present, vision-based hand gesture recognition is very important in human-robot interaction (HRI). This non-contact method enables natural and friendly interaction between people and robots. Aiming at this technology, a two-stream CNN framework (2S-CNN) is proposed to recognize the American sign language (ASL) hand gestures based on multimodal (RGB and depth) data fusion. Firstly, the hand gesture data is enhanced to remove the influence of background and noise. Secondly, hand gesture RGB and depth features are extracted for hand gesture recognition using CNNs on two streams, respectively. Finally, a fusion layer is designed for fusing the recognition results of the two streams. This method utilizes multimodal data to increase the recognition accuracy of the ASL hand gestures. The experiments prove that the recognition accuracy of 2S-CNN can reach 92.08% on ASL fingerspelling database and is higher than that of baseline methods.

This paper presents a new way to teach a robot certain motions remotely from human demonstrator. The human and robot interface is built using a Kinect sensor which is connected directly to a remote computer that runs on processing software. The Cartesian coordinates is extracted, converted into joint angles and sent to the workstation for the control of the Sawyer robot. Kinesthetic teaching was used to correct the reproduced demonstrations while only valid resolved joint angles are recorded to ensure consistence in the sent data. The recorded dataset is encoded using GMM while GMR was employed to extract and reproduce generalised trajectory with respect to the associated time-step. To evaluate the proposed approach, an experiment for a robot to follow a human arm motion was performed. This proposed approach could help non-expert users to teach a robot how to perform assembling task in more human like ways.

No abstract

In this paper, we considered the task of the robot learning low-level trajectory task in a novel clustered constraint environment.We propose a novel adaptive trajectory algorithm used to generate the necessary trajectory which satisfies the constraint of avoiding collision with an obstacle. Our approach is based on Gaussian mixture model which decomposes the trajectory into several ellipses since the isoline of a single Gaussian model is also an ellipse. Moreover, we employed the principle of the artificial potential field to modify the direction of the motion in the presence of obstacles. Since our approach is based on the underlying reactive skill dynamics, it does not share the same disadvantages as approaches which assume both the model of the task trajectory and the response from the obstacle should be learned from the demonstrations.