Introduction
2015; Wiley; Volume: 32; Issue: 3 Linguagem: Inglês
10.1002/rob.21600
ISSN1556-4967
Autores Tópico(s)Biomedical and Engineering Education
ResumoThe Defense Advanced Research Projects Agency (DARPA) Robotics Challenge (DRC) trials were held on December 20-21, 2013, at the Homestead Miami Speedway in Homestead, Florida, in the United States. The DRC Trials are the first in a planned series of two DARPA-sponsored competitions designed to advance the capabilities of humanoid robots; a second competition (the DRC Finals) will be held on June 5-6, 2015, in Pomona, California. The DRC was motivated by a goal of “generat[ing] groundbreaking research and development in hardware and software that will enable future robots, in tandem with human counterparts, to perform the most hazardous activities in disaster zones, thus reducing casualties and saving lives.” Specific motivation was provided by the inability of the robotic community to deliver effective assistance during the Fukushima nuclear disaster of March, 2011. 16 teams from around the world competed in the DRC Trials, with each team associated with one of four development tracks: A, B, C, or D. Track A teams received funding from DARPA to develop both a unique humanoid robot platform and associated software suite. The resulting hardware platforms included bipedal, quadrupedal, and tracked systems. Software suites typically included modules for locomotion, manipulation, state estimation, perception, and tele-operation, among other components. Track B teams received DARPA funding to develop only a software suite to control a DARPA-provided humanoid robot. The DARPA-provided robot was a bipedal robot dubbed “Atlas,” which was developed by Boston Dynamics. Finally, Track C and D teams were not initially funded by DARPA; Track C teams focused solely on software while Track D teams focused on both hardware and software development. Competing teams were scored on their ability to successfully complete intermediate waypoints within each task, within a specific time window. The results of the DRC Trials were both encouraging and sobering. While competing teams demonstrated impressive flashes of performance in the various tasks, and clearly advanced the state-of-the-art in humanoid robotics research, the robots also frequently stumbled, paused, or completely failed to complete the task at hand. The DRC Trials showed that while substantial progress has been made toward the goal of fielding robust, capable humanoid robots, much work remains to be done. The winner of the DRC Trials was a Track A robot developed by Schaft. The Schaft robot was one of the few humanoids to rely on electromagnetic (instead of hydraulic) drive actuators, and displayed an impressive combination of strength and dexterity, resulting in a system that scored a total of 27 points during the trials, easily outdistancing the second-place competitor, IHMC Robotics. The top five finishers were rounded out by Tartan Rescue, Team MIT, and Robosimian. This special issue presents 12 papers from teams that competed in the DRC Trials in two parts, the first in March 2015 and the second in May 2015. All papers were subjected to the standard Journal of Field Robotics peer review process. The paper entitled, “Team THOR's Entry in the DARPA Robotics Challenge Trials 2013,” by Yi et al. describes the technical approach, hardware design and software algorithms employed by Team THOR in the DRC Trials. Modular software and hardware components were employed to enable efficient and cost-effective parallel development. The robot hardware consists of standardized, general purpose actuators and structural components, allowing for efficient maintenance, quick reconfiguration, and a lower build cost compared to other humanoid robots. The software consists of a hybrid locomotion engine, hierarchical arm controller, and platform-independent remote operator interface. The paper “A general-purpose system for teleoperation of the DRC-HUBO humanoid robot,” by Zucker et al. presents a general system focused on addressing three events of the DRC Trials: debris clearing, door opening, and wall breaching. The hardware platform described in the paper is the DRC-HUBO, a redesigned model of the HUBO2+ humanoid robot. The system described in the paper allowed a trio of operators to coordinate a 32 degree-of-freedom robot during a variety of complex mobile manipulation tasks using a single, unified approach. The paper also describes qualitative analysis of lessons learned from the Trials. “Human-Robot Teaming for Rescue Missions: Team ViGIR's Approach to the 2013 DARPA Robotics Challenge Trials” by Kohlbrecher et al. describes Team ViGIR's software development effort for the DRC Trials. The paper outlines the design of an operator control station that allowed multiple operators to request and share information as needed to maintain situational awareness under bandwidth constraints, while directing the robot to perform tasks with most planning and control taking place onboard the robot. Several open source libraries were employed, both for onboard software and operator control station design. The paper concludes with a number of lessons learned from the Trials experience. The paper by Paine et al. entitled “Actuator Control for the NASA-JSC Valkyrie Humanoid Robot: A Decoupled Dynamics Approach for Torque Control of Series Elastic Robots” focuses on control of the Valkyrie robot's series elastic actuators, including description of the control architecture, controller design, and hardware implementation. A decentralized approach is described for controlling Valkyrie's many degrees of freedom. The paper presents the application of this control approach to the Valkyrie robot, which was developed for the DRC Trials. “Valkyrie: NASA's First Bipedal Humanoid Robot” by Radford et al. describes Valkyrie, a 44 degree of freedom, series elastic actuator-based robot that boasts over 18 years of robotics humanoid design heritage. A system overview is first presented that details Valkyrie's mechatronic subsystems. Next, the software and control architectures are highlighted, along with the operator interface and details of the instantiations. Finally, closing remarks are made about the competition and an outlook of future work is outlined. “Analysis of Human-Robot Interaction at the DARPA Robotics Challenge Trials,” by Yanco et al. describes a study of human-robot interaction that was based on direct observation of the participating teams, both from the field and in the control room. The study presents many performance metrics, and categorizes interaction methods according to number of operators, control methods, and interface automation. Each team's interaction methods were compared to their performance, and correlations were analyzed to understand why some teams ranked higher than others. We hope that the papers collected here will be of interest to both roboticists and a wider audience of readers who are interested in learning about the state of the art in humanoid robotics technology. We fully expect that the promising results demonstrated at the Trials will be far surpassed at the upcoming DRC Finals, and will enable a new generation of robotic systems that can accomplish difficult tasks in important real-world scenarios. We would also like to express our gratitude to the many individuals who served as reviewers of these papers, typically through several iterations, and helped ensure their high quality. This special issue is dedicated to the memory of Professor Seth Teller: colleague, friend, and roboticist extraordinaire.
Referência(s)