Stephen G. McGill

Team THOR's Entry in the DARPA Robotics Challenge Trials 2013

This paper describes the technical approach, hardware design, and software algorithms that have been used by Team THOR in the DARPA Robotics Challenge DRC Trials 2013 competition. To overcome big hurdles such as a short development time and limited budget, we focused on forming modular components-in both hardware and software-to allow for efficient and cost-effective parallel development. The hardware of THOR-OP Tactical Hazardous Operations Robot-Open Platform consists of standardized, advanced actuators and structural components. These aspects allowed for efficient maintenance, quick reconfiguration, and most importantly, a relatively low build cost. We also pursued modularity in the software, which consisted of a hybrid locomotion engine, a hierarchical arm controller, and a platform-independent remote operator interface. These modules yielded multiple control options with different levels of autonomy to suit various situations. The flexible software architecture allowed rapid development, quick migration to hardware changes, and multiple parallel control options. These systems were validated at the DRC Trials, where THOR-OP performed well against other robots and successfully acquired finalist status.

Team THOR's adaptive autonomy for disaster response humanoids

This paper describes Team THORs approach to sliding autonomy in manipulation and full body control of a disaster response robot for the 2015 DARPA Robotics Challenge (DRC) Finals. Under the duress of unpredictable bandwidth constraints, autonomous behaviors become critical for reducing response time and dealing with dynamic disturbances. However, the nature of disaster response presents situations where fine grained and intricate teleoperation remain the only safe method of operation. We present sets of algorithms that gracefully switch among high level autonomous behaviors and low levels of teleoperated control. Manipulation algorithms interact in a hierarchical fashion within a state machine, transitioning between states autonomously or through human intervention. Similarly, the balancing controller scales from low degree of freedom walking to full body motions. To validate our methods, we show results from attempts at the DRC Finals and in our preparation for it.

Cooperative humanoid stretcher manipulation and locomotion

In this paper and accompanying video, we demonstrate a humanoid robotic system where two humanoids are physically constrained by the cooperative manipulation of a rigid stretcher body. The humanoids walk together as a virtual quadruped, showing how principles of quadruped gait generation can be applied to cooperating humanoids. In the experiments, two DARwIn-OP humanoid robots estimate the position and orientation of the stretcher, in order to approach it and pick it up. Once they are standing with the stretcher in their hands, they walk by synchronizing their gaits to simulate a virtual quadruped. By systemically tuning the underlying gait parameters, we show how the cooperative humanoids can carry the stretcher in the most stable fashion.

Modular low-cost humanoid platform for disaster response

Developing a reliable humanoid robot that operates in uncharted real-world environments is a huge challenge for both hardware and software. Commensurate with the technology hurdles, the amount of time and money required can also be prohibitive barriers. This paper describes Team THORs approach to overcoming such barriers for the 2013 DARPA Robotics Challenge (DRC) Trials. We focused on forming modular components - in both hardware and software - to allow for efficient and cost effective parallel development. The robotic hardware consists of standardized and general purpose actuators and structural components. These allowed us to successfully build the robot from scratch in a very short development period, modify configurations easily and perform quick field repair. Our modular software framework consists of a hybrid locomotion controller, a hierarchical arm controller and a platform-independent operator interface. These modules helped us to keep up with hardware changes easily and to have multiple control options to suit various situations. We validated our approach at the DRC Trials where we fared very well against robots many times more expensive.

From Reactive to Cognitive Agents: Extending Reinforcement Learning to Generate Symbolic Knowledge Bases

A new methodology for knowledge-based agents to learn from interactions with their environment is presented in this paper. This approach combines Reinforcement Learning and Knowledge-Based Systems. A Q-Learning algorithm obtains the optimal policy, which is automatically coded into a symbolic rule base, using first-order logic as knowledge representation formalism. The knowledge base was embedded in an omnidirectional mobile robot, making it able to navigate autonomously in unpredictable environments with obstacles using the same knowledge base. Additionally, a method of space abstraction based in human reasoning was formalized to reduce the number of complex environment states and to accelerate the learning. The experimental results of autonomous navigation executed by the real robot are also presented here.

Active stabilization of a humanoid robot for real-time imitation of a human operator

Imitating the motion of a human operator is an intuitive and efficient way to make humanoid robots perform complex, human-like behaviors. With the help of recently introduced affordable and real-time depth sensors, the real time imitation of human behavior has become more feasible. However, due to their small footprint and high center of mass, humanoid robots are not inherently stable. The momentum generated by dynamic upper body movements can induce instabilities that are often large enough to make the robot fall down. In this work, we describe a motion controller for a humanoid robot where the upper body is controlled in real time to imitate a human teacher, and the lower body is reactively stabilized based on the current measured state of the robot. Instead of relying on the accuracy of robot dynamics, we use biomechanically motivated push recovery controllers to stabilize the robot against unknown perturbations that include possible impacts. We demonstrate our approach experimentally on a small humanoid robot platform.

Extensions of a RoboCup Soccer Software Framework

The RoboCup soccer leagues have greatly benefitted Team DARwIn and the UPennalizers. The stiff competition has hardened our code into a robust framework, and the community has allowed it to flourish as an open source project used by many teams. Working with the open source DARwIn-OP hardware allows even more clairvoyance into the inner workings on the low level code the builds to state machines. We show how our codebase performs in the Webots simulation and on the Open Source DARwIn-OP platform. From these beginnings, we wish to apply our codebase to new scenarios for humanoids including human robot interaction and manipulation tasks. Many of these scenarios are explored by other RoboCup leagues including @Home and Rescue, where we see a new avenue for our codebase. New human robot interaction features are described in our framework, and example performances are demonstrated. Finally, we describe added standards compliance and open source tool usage that will give our codebase more accessibility.

Hierarchical Motion Control for a Team of Humanoid Soccer Robots

Robot soccer has become an effective benchmarking problem for robotics research as it requires many aspects of robotics including perception, self localization, motion planning and distributed coordination to work in uncertain and adversarial environments. Especially with humanoid robots that lack inherent stability, a capable and robust motion controller is crucial for generating walking and kicking motions without losing balance. In this paper, we describe the details of a motion controller to control a team of humanoid soccer robots, which consists of a hierarchy of controllers with different time frames and abstraction levels. A low level controller governs the real time control of each joint angle, either using target joint angles or target endpoint transforms. A mid-level controller handles bipedal locomotion and balancing of the robot. A high level controller decides the long term behavior of the robot, and finally the team level controller coordinates the behavior of a group of robots by means of asynchronous communication between the robots. The suggested motion system has been successfully used by many humanoid robot teams at the RoboCup international robot soccer competitions, which has awarded us five successful championships in a row.

Low dimensional human preference tracking for motion optimization

Motion planning for high degree of freedom (DOF) robots is not an easy task, and often requires optimization in a high dimensional space. Still, a generic motion planner using a single cost function for optimization may not be optimal over a number of different tasks with various task specific constraints. In this paper, we present a motion planning system that utilizes both easy to communicate human preferences and dimensionality reduction to handle these issues. Joint trajectories with human preference costs are projected into the null space of the task space, which helps make the resulting optimization simpler and more reliable. In addition, we apply the dimensionality reduction for the optimization, which significantly lowers the computational load. The suggested controller has been successfully used in the DARPA Robotics Challenge (DRC) Finals to handle a number of manipulation tasks.

THOR-OP humanoid robot for DARPA Robotics Challenge Trials 2013

This paper describes the hardware design and motion control algorithms that have been used by Team THOR in the DARPA Robotics Challenge (DRC) Trials 2013 competition. The robotic hardware we use, the THOR-OP robot, consists of standardized and general purpose actuators and structural components, which greatly reduce the build and reconfiguration time and allows for quick field repair capability. Our software framework is also composed of fully modular function modules. This modular structure helps us to keep up easily with hardware changes and to have multiple control options to suit various situations. We validated our approach at the DRC Trials where we fared well against other robots many times more expensive and acquired the finalist status.