In response to the outbreak of highly infectious diseases, such as Ebola (2015) and Zika (2016), a Tele-Robotic Intelligent Nursing Assistant (TRINA) was developed to assist healthcare workers in routine patient-caring tasks, handling of contaminated materials and protective gear. At WPI, we further improve TRINA’s level of automation and user interface, for patient-caring and general-purpose assistive tasks, to provide appropriate cognitive and physical augmentation to diverse healthcare workers.
Here are our complete and on-going projects on TRINA platform:
- Shared autonomous perception-action coordination
- 2019/03-present, shared-autonomous camera selection and control for tele-nursing robot
- 2018/06-2019/03, perception-action coordination in VR telepresence (complete)
- Intuitive and user-adaptive teleoperation interface
- 2019/03-present, assessing teleoperation fatigue using OpenSim human model
- 2018/09-2019/03, assessing the physical fatigue in teleoperation via motion mapping: a pilot study (complete)
- 2017/01-2018/04, teleoperation interfaces via whole-body human motion mapping (complete)
- Learning and planning in human-robot collaboration
- 2019/02-present, Learning and planning with parameterized symbol
- 2018/01-2019/01, High-level learning and planning for loco-manipulation (complete)
- 2017/09-2018/09, Predicting object transfer point for fluent human-robot handover (complete)
- Tele-Robotic Intelligent Nursing Assistant (TRINA) at Duke University
- 2015/01-2016/03, System integration and performance evaluation (complete)
- 2016/01-2016/05, Bidirectional telepresence for robot-mediated handover (complete)
High-fidelity computational human models provide a safe and cost-efficient method for the study of driver experience in vehicle maneuvers and for the validation of vehicle design. Compared to passive human model, an active human model that can reproduce the decision-making, as well as vehicle maneuver motion planning and control will be able to support more realistic simulation of human-vehicle interaction. To this end, we propose a integrated human-vehicle interaction simulation framework that can learn the motion primitives of vehicle maneuver motions from human drivers, and use them to compose natural and contextual driving motions in simulation. Such active computational human model can also be used to analyze human performance and efforts in the interactions with assistive robots and environments, and evaluate design ergonomics and compatibility.
Here are our active projects for computational modeling of human-vehicle interaction:
- 2018/10-present, Force anticipation in pushing and pulling tasks
- 2018/05-present, Human Model-based Evaluations of Overtaking Performance and Driver Control Workloads
- 2017/04-2018/03, Learning motion primitive library for vehicle maneuver motion coordination (complete)
- 2017/04-2018/04, A novel simulation framework for human-vehicle interaction (complete)
Before WPI: Upper Limb Exoskeleton for Stroke Rehabilitation
The motion compatibility of a wearable robot system and its operators is critical to exoskeletons for power augmentation and rehabilitation. Ideally, the motion of a wearable robot system should be dynamically transparent to a healthy operator, sensitively responsible to the voluntary and involuntary motions of its operator. When used for robot-assisted stroke rehabilitation and/or power augmentation, a wearable robot system is expected to assist to the operator’s motor skills and correct abnormal arm motions resulting from motor disabilities. To this end, this research efforts studied the arm motions in reaching and grasping tasks, to (1) reveal the underlying control strategy of the human motor system, and to (2) inspire the design and motion control of arm-compatible/arm-like robotic manipulators for overall performance improvement.
Here are the completed projects on bio-inspired motion planning for upper limb exoskeletons:
- Bio-inspired kinematic redundancy resolutions
- Synthesizing motion control criteria for arm posture prediction
- Arm-Hand Coordination in Reach-to-Grasp Motions
- Robot-assisted stroke rehabilitation using inter-arm teleoperation
Before WPI: Multi-arm Surgical Robot System
This research effort aims to optimize the mechanical design, configuration and motion coordination of the Raven IV surgical robot. For best manipulation dexterity in the common workspace of the four surgical robot arms, we proposed a performance index that considers link rigidity, operation area, and manipulability for all possible robot arm configurations. We optimized the link lengths, robot base orientation and port spacing.
In addition, my research on the arm-hand coordination of wearable robots also provided guidelines for the coordination of macro- and micro-mechanism of robotic manipulators. For a surgical robot arm, the macro-mechanism is the supporting robot arm, while the micro-mechanism is the attached surgical tool for operation. Similar to the human arm and hand, the macro-mechanism positions the hand for best manipulability while the micro-mechanism accomplishes the task with best accuracy. In hand-arm coordination, I especially investigated the coordination of arm posture and supination-pronation of the forearm. My results suggested that the synergy between the macro- and micro-mechanisms gradually breaks down as the either of them get close to their joint limits.
Projects along this line include:
- Design and configuration optimization for manipulation dexterity
- Bio-inspired motion coordination of macro- and micro-structures
Before WPI: Multi-user Cooperative Manipulation via Networked Haptics
Networked haptics enables cooperative manipulation of shared virtual objects and direction interactions with multiple remote users. Control architectures are necessary for synchronizing the distributed copies of the share virtual objects, as well as maintaining the contact stability of multi-agent interaction. To this end, we propose to use remote dynamic proxies with wave-based controllers to represent users in remote virtual environment. We found the remote dynamic proxies can render smooth motion and force feedback in the presence of long and time-varying time delay, and provide better user state synchronization.