Design and control of the rapid legged platform GAZELLE
Introduction
Man-made devises are often motivated by the design and mechanism of men themselves. A humanoid robot is a clear example of this. To improve performance and realize more adaptability in a human working environment, robots’ configurations have been made very similar to humans. In the DARPA Robotics Challenge finals in 2015 (DRC finals), challenging tasks have been given assuming a radiation disaster in an environment where humans work [1]. Many bipedal humanoid robots in the DRC finals showed state-of-the-art performances [2], [3], [4], [5], [6], [7]. Most tools and environments are fit for humans; hence, the development of a human-like bipedal locomotion is both necessary and important as an alternative for wheel-based robots.
Bipedal platforms have been developed and researched upon from decades ago. In many universities, bipedal platforms have been developed for locomotion study. Waseda University has performed research on the development of the WABIAN series [8]. They implemented a 6-degree-of-freedom (DOF) parallel manipulator to a leg in a WL-16- and WABIAN-RIV-mounted stereo vision system. After that, the joint position-controlled humanoid, WABIAN-2, was developed, which can walk with a stretched knee [9]. Oregon State University also developed a spring-loaded inverted pendulum model for model-based walking with its robot, ATRIAS [10]. Recently the team, now working as Agility Robotics, has developed an upgraded platform, called Cassie [11]. Their series of platforms use a series elastic actuator for the torque control of joints. Technical University of Munich developed Johnnie [12] and Lola [13]. Lola consists of a 7-DOF leg with shock-absorbing feet. It reduces its weight through optimal design and uses links to actuate the joints. KAIST began developing the KHR series [14,15], and is now working on the HUBO series [5,16].
Industries were also developing unique humanoid robots. Honda developed its ASIMO [17]. Toyota worked on a partner robot [18]. Samsung developed Roboray [19]. All these robots have shown impressive performances. Research facilities also took part in the humanoid development. AIST has maintained and updated its HRP series humanoid robot [20], [21], [22]. KIST developed MAHRU-R [23]. DLR developed the torque-controlled robot, TORO [24]. IIT developed the compliant humanoid, COMAN [25]. Different research facilities have also used humanoid platforms to develop stabilizing control, path planning, and motion generation.
Research with regard to the performance of humanoid parts, such as motor cooling, has been performed. Urata et al. [26] developed a liquid-cooling module for motor heat dissipation. Paine and Sentis [27] analyzed a motor's thermal dynamics to assess the liquid cooling efficiency. Jung et al. [5] suggested a robot design for motor heat dissipation using body frame contact and cooling fin.
The current study presents the development of a lightweight and fast biped platform. We overview the basic specifications of GAZELLE and briefly describe its electronic, software, and sensory systems. The electronic and software systems are constructed based on the system of DRCHUBO+ [5], which won the first prize on the DRC finals. It showed a stable performance during the competition; hence, the reliability of the electronic and software systems has been verified. In the mechanical design part, we present a linked design of the ankle structure, which is 2-RSU (Rotation-Spherical-Universal) parallel manipulator. The kinematics and Jacobian of the 2-RSU manipulator are described, and the torque control of the ankle joint is realized using a virtual work principle. We suggest a novel method of the heat dissipater design for a cylinder-shaped motor to maximize the motor performance. This air-cooling unit performs as well as the water-cooling unit, and is extremely light and energy efficient. We conduct a walking experiment to verify the performance of the overall platform.
The rest of this study is organized as follows: Section 2 presents a brief explanation of the actuator and introduces the electronic, software, and sensory systems; Section 3 describes the special characteristics of our mechanical design; Section 4 describes the walking algorithm of GAZELLE and verifies the performance of the overall platform through the experiments; and finally, Section 5 concludes the paper.
Section snippets
Bipedal platform: GAZELLE
GAZELLE is a 13-DOF bipedal platform that is able to move fast and robustly actuated. It is designed for a bipedal walking experiment in various environments, such as uneven terrain and external disturbances (Fig. 1). Table 1 shows its overall specification. Fig. 2 presents its kinematic structure and dimensions. Although GAZELLE has a parallel manipulator structure on the ankle, it can be considered coinciding ankle pitch and roll axes.
Joint actuator
The actuator system consists of a motor and a harmonic
Hip and knee joints
The hip roll and pitch axes were separated to ensure the hip joint's angle range (Fig. 6(a)). By separating the hip roll axis and the pitch axis, both joints can simultaneously have a wide joint range. This leg configuration still has an analytic inverse kinematics solution because it has three successive parallel axes (i.e., hip pitch, knee, and ankle pitch) [28]. Thanks to the wide range of joint angles, we can have a motion that requires the flexibility of a hip joint, such as walking in one
Walking pattern and stabilization control
The basic components of walking are the walking patterns and the controllers that stabilize it. Walking patterns are generated based on the robot model, while controllers are constructed based on the sensor feedback. This section will briefly introduce the walking pattern generation method and the stabilization controller.
Conclusion
In this study, we designed the lower-body platform, GAZELLE, which was lightweight and capable of a high power output. Lightweight actuators and an efficient structure design reduced its weight. It can also reduce the moment of inertia of the lower body using the link structure. By using the link structure for ankle joint, the stiffness of the joint is lowered and the joint backlash would be occurred. However, in the case of the ankle joint, the joint stiffness and backlash do not affect
CRediT authorship contribution statement
Hyobin Jeong: Conceptualization, Methodology, Software, Writing - original draft. KangKyu Lee: Data curation, Visualization, Validation. Wooshik Kim: Writing - original draft, Investigation. Inho Lee: Writing - review & editing, Supervision. Jun-Ho Oh: Conceptualization, Validation, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgment
This work was supported by the ‘Development of core technologies and a standard platform for humanoid robot [10060103]’, which is a project from the Ministry of Trade, Industry and Energy (MOTIE) of the Republic of Korea.
Hyobin Jeong received the B.S., M.S., and Ph.D. degrees in mechanical engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea, in 2012, 2014, and 2019, respectively. He researched for the project of development for humanoid robots: HUBO2 and DRC-HUBO. His research interests include development of six-axis force/torque sensor, mechanical design of bipedal robot, and bipedal walking stabilization control algorithm. Dr. Jeong was a member of TEAM KAIST
References (42)
- et al.
Balance recovery through model predictive control based on capture point dynamics for biped walking robot
Rob Auton Syst
(Jul. 2018) The DARPA robotics challenge finals: results and perspectives
J F Robot
(Mar. 2017)Team SNU's control strategies for enhancing a robot's capability: lessons from the 2015 DARPA robotics challenge finals
J F Robot
(2017)Team IHMC's lessons learned from the DARPA robotics challenge trials
J F Robot
(Mar. 2015)Team WPI-CMU: achieving reliable humanoid behavior in the DARPA robotics challenge
J F Robot
(Mar. 2017)- et al.
Development of the humanoid disaster response platform DRC-HUBO+
IEEE Trans Robot
(Feb. 2018) WALK-MAN humanoid lower body design optimization for enhanced physical performance
Proc - IEEE Int Conf Robot Autom
(June 2016)Development of humanoid robot system for disaster response through team NEDO-JSK's approach to DARPA robotics challenge finals
- et al.
Biped walking robots created at Waseda university: WL and WABIAN family
Philos Trans R Soc A Math Phys Eng Sci
(Jan. 2007) Development of a new humanoid robot WABIAN-2
ATRIAS: design and validation of a tether-free 3D-capable spring-mass bipedal robot
Int J Rob Res
Mechanical design for robot locomotion
Sensors and control concept of a biped robot
IEEE Trans Ind Electron
System design and control of anthropomorphic walking robot LOLA
IEEE/ASME Trans Mechatron
Walking control of the humanoid platform KHR-1 based on torque feedback control
System design and dynamic walking of humanoid robot KHR-2
Mechanical design of humanoid robot platform KHR-3 (KAIST humanoid robot - 3: HUBO)
The development of Honda humanoid robot
Fast running experiments involving a humanoid robot
Development of the lower limbs for a humanoid robot
Design and experiments of advanced leg module (HRP-2 L) for humanoid robot (HRP-2) development
Cited by (12)
Slope walking of humanoid robot without IMU sensor on an unknown slope
2022, Robotics and Autonomous SystemsCitation Excerpt :Additionally, they proposed a CoM damping controller and foot force/torque control to stabilize the CoM using the calculated CoM, ZMP, and CP. The humanoid robot Gazelle walked stably using the proposed controller on uneven terrain with a changing slope angle [15]. Additionally, there have been various studies on state-estimation methods for robots [16–22].
Seamless Reaction Strategy for Bipedal Locomotion Exploiting Real-Time Nonlinear Model Predictive Control
2023, IEEE Robotics and Automation LettersArtificial neural network-based ground reaction force estimation and learning for dynamic-legged robot systems
2023, PeerJ Computer ScienceLegway: Design and Development of a Transformable Wheel-Leg Hybrid robot
2023, IEEE-RAS International Conference on Humanoid RobotsDesign of New Drive Mechanism for Dynamic Movement of Humanoid Robot Legs
2023, IEEE-RAS International Conference on Humanoid Robots
Hyobin Jeong received the B.S., M.S., and Ph.D. degrees in mechanical engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea, in 2012, 2014, and 2019, respectively. He researched for the project of development for humanoid robots: HUBO2 and DRC-HUBO. His research interests include development of six-axis force/torque sensor, mechanical design of bipedal robot, and bipedal walking stabilization control algorithm. Dr. Jeong was a member of TEAM KAIST that won the first place in the DARPA Robotics Challenge Final in 2015.
KangKyu Lee received his B.S. degree in Mechanical Engineering from Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea, and M.S. in Mechanical Engineering from KAIST, in 2013 and 2015, respectively. Since 2015, he is currently pursuing the Ph.D. degree in Mechanical engineering at the KAIST and working on the project of development for humanoid robots: DRC-HUBO. His research interests include whole body control, floating-base dynamics, actuation system and humanoid design. And he is a member of TEAM KAIST that won the 1st place in DARPA Robotics Challenge Final 2015. ([email protected])
Wooshik Kim received his B.S. degree in Mechanical Engineering from Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea in 2017. Since 2017, he is currently pursuing Master's degree in Mechanical engineering at Carnegie Mellon University in Pittsburgh Pennsylvania, United States of America. He is currently working on path planning for manipulating continuously deformable environments like sand and developing simple model of their interaction. He worked on developing lower level control—kinematics and torque control—of HUBO's new lower body Gazelle for undergraduate research. ([email protected])
Inho Lee received his B.S. degree, M.S. and Ph.D degree in Mechanical Engineering from Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea, in 2009, 2011 and 2016, respectively. He was working on the project of development for humanoid robots: HUBO, HUBO2 and DRC-HUBO. In 2015, He was a team member of TEAM KAIST which is the winning team for DARPA Robotics Challenge. He is currently working in Institute for Human and Machine Cognition Robotics Lab (IHMC) and his research interests include motion planning, quadruped and bipedal walking and stabilization control for a humanoid robot, manipulation, sensors, actuators and application of micro-processor. ([email protected])
Jun-Ho Oh received his B.S. and M.S. degrees in Mechanical Engineering from Yonsei University, Seoul, South Korea, and has Ph.D. degree in Mechanical Engineering from University of California, Berkeley, in 1977, 1979, and 1985 respectively. He was a Researcher with the Korea Atomic Energy Research Institute, from 1979 to 1981. Since 1985, he has been with the Department of Mechanical Engineering, KAIST, where he is currently a significant professor and a director of Humanoid Robot Research Center. And he has been a vice president of KAIST since 2013. He was a Visiting Research Scientist in the University of Texas Austin, from 1996 to 1997. His research interests include humanoid robots, adaptive control, intelligent control, nonlinear control, biomechanics, sensors, actuators, and application of micro- processor. Dr. Oh is a member of the IEEE, KSME, KSPE and ICASE. ([email protected])