Official websites use .gov
A .gov website belongs to an official government organization in the United States.
Secure .gov websites use HTTPS
A lock (
) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.
Summary
Mobile Robot Systems, which include autonomous mobile robots, quadrupeds, humanoids, mobile manipulators, and wearable robots (active, passive and hybrid), are expanding their capabilities well beyond the traditional roles as they attain greater onboard intelligence and control. To improve industrial processes, companies are requiring robots to become more adaptable, faster, agile and highly accurate in dynamic environments. One example is the ability to move the manipulator arm to different workstations as needed without requiring the robot to relocalize. The mobile manipulator would be able to coordinate its base with its arm to perform continuous operations in presence of obstacles and perturbations. The project will focus on metrology for integrated AI capabilities and industry-driven standards in the context of dynamic learning and adaptation, while maintaining assurance the mobile robot system can meet operational constraints. The project will also be developing metrics, testbed infrastructure, and measurement systems to validate performance test methods for mobile robots.
Wearable robots are becoming more advanced with increasingly adaptable with more dynamic and accurate control, to prevent worker fatigue, improve task generalizability. Wearable robots can improve worker generalizability by augmenting a worker’s physical capability to perform additional job functions. Some manufacturing companies are even requiring wearable robots as personal protective equipment. The project will develop metrology for advanced learning capabilities of wearable robots by instrumenting new sensors into the testbed infrastructure to advance the test methods and algorithms for benchmarking how active and hybrid wearable robots learn and adapt to users, tasks, and the environment. The knowledge learned will inform ASTM F48 standards development.
Description
Objective
Develop measurement science for intelligent industrial mobility systems, including mobile and wearable robotic systems.
Technical Idea
This project will develop common reference measurements and evaluation techniques for mobile robot systems with integrated learning (AI) capabilities. These systems include integrated autonomous mobile robot systems with simultaneous localization and mapping capabilities, as well as active exoskeletons with integrated learning of user kinematics and kinetics.
Building on state-of-the-art industrial localization and closed-loop control robots commercially available in the market today, the project’s next phase is to research and develop measurement science for learning and intelligence capabilities for mobile robot systems. Common issues encountered in prior localization tests conducted include slippage, uneven flooring, dynamic obstacles, vibrations, heavy and dynamic loads, and forced delocalization (after the robot builds its map, the operator changes the physical orientation and or location of the robot). Machine Learning/Artificial Intelligence (ML/AI) capabilities for mobile robot systems promise to better adapt to these scenarios, and new measurement science and standard test methods are needed to validate the ML/AI capabilities.
Considering significant advances in mobile robotic systems intelligence and autonomy, the project approach will: 1) measure performance of mobile robotic system control and onboard sensing using manual mapping (joystick navigation to map prior to autonomous control), 2) measure performance of mobile robotic system control and onboard sensing using learning (exploring the environment autonomously to develop a map), and 3) integrate external optical tracking with mobile robotic systems to evaluate localization in dynamic, unstructured environments (e.g., discrete and continuous workpiece movement, obstacle avoidance).
Building on prior research methodology for designing metrics and test methods for binary (on/off) passive wearable robots will be advanced into metrics and test methods for active and hybrid systems – i.e., the current industry request for systems that will change the amount of assistance needed as required by the end user (continuously changing force/pressure wearable robot control).
A more responsive wearable robot with closed loop sensory control would enable continual assistance and situational awareness (e.g., lift position, predictive muscle activity, endurance) when needed by the user. Wearable robot manufacturers are already marketing these advanced systems (hybrid, active) yet test methods are lacking for these new systems. Evaluation of adaptive wearable robots requires continuous, synchronous measurements of both human and wearable robots including but not limited to, joint angles, velocities, accelerations and torques. The project will utilize both advanced optical tracking capabilities and low-cost hardware augmented with ML/AI to compute measurements of human-wearable robot performance. The optical tracking will be used as ground truth to evaluate the uncertainty of the low cost, fieldable measurements.
Research Plan
New measurement methods are needed to evaluate the performance of dynamic localization of mobile robotic systems and wearable robots with continuous, adaptive control and situational awareness. The research plan will develop new metrics, test methods, artifacts, and datasets that measure the performance characteristics for these devices. Position estimation and tracking the kinematics of mobile manipulators, mobile robots, and wearable robots are essential to quantifying a robot’s capabilities. Utilizing a well-established and highly successful partnership with ASTM F45/F48 we will draft novel test methodologies, test artifact designs, and experimental results for further development and balloting. The project will develop concepts for characterizing the levels of intelligence and autonomy-based capabilities for mobile robotic systems as guidance for the research and user community and as input to standards roadmaps.
Towards advanced mobile robot systems capable of adapting and learning new tasks, the plan for the first year is to develop benchmarking metrics and novel measurement methods leveraging AI to enhance metrology and evaluation of mobile robot system stability, adaptability, task efficiency, task completion quality, and task assurance considerations. Our plan is to integrate a state of the art, commercially available, open source mobile manipulator. We will characterize the adaptability and learning capabilities and explore avenues to augmenting the capabilities by instrumenting additional onboard sensing and extending potential learning algorithms based on in-situ measurements providing the robot additional context for localization, planning and navigation in year 2.
In year 3, with additional contextual information, the plan is to integrate two heterogeneous mobile robot, such as a mobile manipulator and an advanced mobile robot platform to collaborate on a material handling task, such as Additive Manufacturing (AM) onboard a docked AMR on the factory floor.
Major Accomplishments
- Completed implementation of a closed-loop mobile manipulator control system for testing performance with workpiece disturbances.
- Completed implementation of performance test methods for small obstacle avoidance, weighted driving and docking, driving and docking on ramps, and load stability of A-UGVs.
- Helped establish ASTM International Subcommittee F45.05 on Grasping and Manipulation
Published Standards
- ASTM F48
- F3581 Standard Test Method for Exoskeleton Use: Hurdles
- F3582 Standard Test Method for Exoskeleton Use: Gaps
- F3583 Standard Test Method for Exoskeleton Use: Beams
- F3584 Standard Test Method for Exoskeleton Use: Obstacle Avoidance: Walking
- F3613 Standard Practice for Recording the Exoskeleton Fit to the User
- F3614 Standard Practice for Recording the Exoskeleton User Information
- ASTM F45
- F3200 Standard Terminology for Robotics, Automation, and Autonomous Systems
- F3218 Standard Practice for Documenting Environmental Conditions for Utilization with A-UGV Test Methods
- F3243 Standard Practice for Implementing Communications Impairments on A-UGV Systems
- F3244 Standard Test Method for Navigation: Defined Area
- F3327 Standard Practice for Recording the A-UGV Test Configuration
- F3499 Standard Test Method for Confirming the Docking Performance of A-UGVs
- F3588 Standard Guide for Set of Objects used with A-UGVs
Datasets
- Continuous mobile manipulator performance experiment 02-01-2022. https://doi.org/10.18434/mds2-3061
- Exoskeleton Performance Data (2021) https://doi.org/10.18434/mds2-2429