Huan Tan

Towards safe robot-human collaboration systems using human pose detection

This paper proposes a human detection-based cognitive system for robots to work in human-existing environment and keep the safety of humans. An integrated system is implemented with perception, recognition, reasoning, decision-making, and action. Without using any traditional safety cages, a vision-based detection system is implemented for robots to monitor the environment and to detect humans. Subsequently, reasoning and decision making enables robots to evaluate the current safety-related situation for humans and provide corresponding safety signals. The decision making is based on maximizing the productivity of the robot in the manipulation process and keep the safety of humans in the environment. The system is implemented with a Baxter humanoid robot and a PowerBot mobile robot. Practical experiments and simulation experiments are carried out to validate our design.

An integrated vision-based robotic manipulation system for sorting surgical tools

In this paper, we introduced a robotic system using a humanoid robot, Baxter research robot, to pick-up surgical tools from a tray and place the tools into different trays according to the types of the surgical tools. The pick-n-place manipulation is integrated with a vision component and a special magnet gripper and governed by a finite state machine. This vision-based manipulation system allows the robot to check which tool is on top of the tools in a tray, to find the grasping points on the tools, to grab the tools using a magnet gripper, and to place them into different trays. Major technologies used in this system include: vision, magnet force control, force feedback, motion trajectory planning, and decision-making. We tested our system in a lab-based environment and the system performance satisfies the requirements of the project.

Robotic Handling of Surgical Instruments in a Cluttered Tray

We developed a unique robotic manipulation system that accurately singulates surgical instruments in a cluttered environment. A novel single-view computer vision algorithm identifies the next instrument to grip from a cluttered pile and a compliant electromagnetic gripper picks up the identified instrument. System is validated through extensive experiments. This research was motivated by the challenges of perioperative process in hospitals today. Current process of instrument counting, sorting, and sterilization is highly labor intensive. Improperly sterilized instruments have resulted in many cases of infections. To address these challenges, an integrated robotic system for sorting instruments in a cluttered tray is designed and implemented. A digital camera is used to capture an image of a cluttered tray. A novel single-view vision algorithm is used to detect the instruments and determine the top instrument. Position and orientation of the top instrument is transferred to a robot. A compliant electromagnetic gripper is developed to complete the gripping. Experiments have demonstrated high success rate of both instrument recognition and manipulation. In the future, error handling needs to be further reinforced under various exceptions for better robustness.

Integration of evolutionary computing and reinforcement learning for robotic imitation learning.

This paper proposes an evolutionary reinforcement learning method by combining Estimation of Distribution Algorithm and Reinforcement Learning. The Reinforcement Learning method in our method is based on Policy Improvement with Path Integrals (PI2). Estimation of Distribution Algorithm is incorporated into this reinforcement learning method to improve the generation of roll outs with certain noises. This method can accelerate the converging of the learning results and improve the overall system performance. Additionally, this method provides a potential solution to integrate the exploratory evolutionary algorithms and the greedy policy learning method. The proposed method is applied in a robotic imitation learning experiment in this paper and the experimental results demonstrate the effectiveness and robustness of our proposed algorithm.

An integrated robotic system for transporting surgical tools in hospitals

The performance of a hospitals sterile processing center (SPC) significantly impacts patient safety and overall productivity. Key to automating this process is to reliably transport instruments throughout the process. In this paper, we detail a robust integrated system for enabling mobile robots to autonomously perform manipulation of assets; specifically, transporting reusable surgical instrument trays in the SPC of a hospital. Our method is based on a cognitive decision making mechanism that plans and coordinates the motions of the robot base and the robot manipulator at specific processing locations. A vision-based manipulator control algorithm was developed for the robot to reliably locate and subsequently pick up surgical tool trays. Further, to compensate for perception and navigation errors, we developed a robust self-aligning end-effector that allows for improved error-tolerance in larger workspaces. We evaluated the developed integrated system using an Adept PowerBot mobile robot equipped with a 6-DOF Schunk PowerCube arm and our customized end-effector in an SPC-like environment. The experiment results validate the effectiveness and robustness of our system for handling surgical instrument trays in tight and constrained environments.

Applying an Extension of Estimation of Distribution Algorithm (EDA) for Mobile Robots to Learn Motion Patterns from Demonstration

This paper proposes a probabilistic evolutionary computing algorithm for robots to learn motion patterns. This algorithm is inspired from Estimation of Distribution Algorithms (EDA). The distribution of chromosomes (not the genes), which have higher fitness values in the configuration space, is estimated in a configuration space. A modified Probabilistic Rapidly growing Random Tree (PRRT)-Connect algorithm is used for searching the configuration space to generate chromosomes which are represented as paths from the starting point to the goal point. Mutation is defined as searching with certain probability outside of the current distribution area (obstacle-free area). This algorithm is applied for robotic learning of motion trajectories through imitation. Simulation and practical experimental results are given in this paper to verify the effectiveness of this algorithm. The major contribution of this paper is proposing an extension of current EDAs, which could be applied for rapid robotic imitation learning.

Human-Supervisory Distributed Robotic System Architecture for Healthcare Operation Automation

This paper proposes a human-supervisory distributed robotic software architecture, which has been applied in a multi-agent robotic system to automate the daily and repeated sterilization process at hospitals of US Department of Veteran Affairs. Each robot is considered as an independent agent to perform assigned tasks with its own capability and coordinate their operations with other robots to ensure that the main process of the work flow to satisfy the overall operation requirements. This layered architecture highlights human factors in the automation work flow to provide a flexible and robust human-knowledge-based supervision and control for safe, reliable, and automated process for healthcare industry. The proposed architecture and the implemented system were tested in a practical project to validate its effectiveness and robustness.

A Behavior Generation Framework for Robots to Learn from Demonstrations

This paper proposes a framework of generating behavior sequences for robots, especially for humanoid robots, to perform complex tasks. This framework provides a method for the robot to generalize common features of demonstrated behaviors, to store the learned behaviors in the memory system, to construct a behavior graph to describe relationships among learned behaviors, to find and assemble a behavior sequence, and to generate similar motion trajectories of basic behaviors when it is placed in a similar but slightly different task-relevant situation. Additionally, we successfully use behavior graph to describe the dynamic relationship among behaviors and we successfully apply shortest path searching methods in behavior sequence generation, which provides a novel solution to associate knowledge representation with behavior generation. Simulation and experiments are carried on a humanoid robot to validate our proposed framework.

An Integrated Robotic System for Autonomous Brake Bleeding in Rail Yards

Current operations in rail yards are dangerous and limited by the operational capabilities of humans being able to perform safely in harsh conditions while maintain high productivity. Such issues call out the need for robust and capable autonomous systems. In this paper, we outline one such autonomous solution for the railroad domain, capable of performing the brake bleeding inspection task in a hump yard. Towards that, we integrated a large form factor mobile robot (the Clearpath Grizzly) with an industrial manipulator arm (Yasakawa Motoman SIA20F) to effectively detect, identify and subsequently manipulate the brake lever under harsh outdoor environments. In this paper, we focus on the system design and the core algorithms necessary for reliable and repeatable system execution. To test our developed solution, we performed extensive field tests in a fully operational rail yard with randomly picked rail cars under day and night-time conditions. The results from the testing are promising and validate the feasibility of deploying an autonomous brake bleeding solution for railyards.