Legged motion control

Dynamically stable gaits for quadruped walking vehicles

For dynamically stable walking in a quadruped robot, a combination of trajectory planning for body and leg positions (feedforward control) and adaptive control using sensory information (feedback control) is crucial. To achieve this, we propose a new control technique for stable trot gait called the 3D sway compensation trajectory. This method employs rotational and translational body motions along a diagonal line between supporting feet, coupled with vertical swing motions of the recovering legs.

Vertical leg motion Stabilization of unsymmetrical trot gait Intermittent Trot Gait

Papers

Study on energy efficiency of quadruped walking vehicles

Although a legged robot offers superior terrain adaptability compared to a wheeled vehicle, its speed is generally quite low. To achieve higher speeds with a legged robot, dynamically stable gaits like running for biped robots and trot or bound gaits for quadruped robots are promising solutions. However, the energy efficiency of dynamically stable gaits is usually lower than that of more stable gaits, such as the crawl gait. We are conducting an experimental study on the energy efficiency of a quadruped walking vehicle. Through experiments using a quadruped walking vehicle named TITAN-VIII, we investigate the energy consumption of two walking patterns for trot gait. The results show that the proposed 3D sway compensation trajectory offers advantages in energy efficiency compared to the original sway compensation trajectory.

Papers

Stable gait control for a biped robot

The sway compensation trajectory for dynamically stable walking, originally developed for a quadruped robot, has been adapted for use in biped robots. This method simplifies the design of stable ZMP (Zero Moment Point) and COG (center of gravity) trajectories, which have traditionally been seen as complex and delicate challenges. The effectiveness of the proposed method has been validated through computer simulations and walking experiments using humanoid robots HOAP-1 and HOAP-2.

Walking experiment Dance by HOAP Omni-directional walking motion
Step climbing Step climbing Dance step by fusing dynamically and statically stable walking

Papers

Straight legged walking for a biped robot

When humans walk, their knee joints stretch periodically. This motion, however, is very difficult for a biped robot due to the limited degrees of freedom (DOFs). We propose a new methodology for generating a straight-legged walking pattern for a biped robot, utilizing the up-and-down motion of the upper body. First, we define two new indices: the Knee Stretch Index (KSI) and the Knee Torque Index (KTI), which indicate how efficiently the knee joints are utilized. Next, the up-and-down motion of the upper body is automatically planned so that these indices are optimized, and straight-legged walking is achieved. The basic idea of the proposed method is:

  1. When a large number of DOFs are required for controlling the Zero Moment Point (ZMP), the robot lowers its body height.
  2. When there are extra DOFs, the body is lifted, and the knee joint is stretched.
By stretching the knee joints, a human-like natural walking motion is obtained. Moreover, energy efficiency is improved since the required torque and energy consumption to support the body weight are reduced at the knee joints.

Straight Legged Walking Straight Legged Walking

Papers

Proactive Human Interface based on Embodied Agents

We are conducting research on "Embodied Proactive Human Interface." This research aims to develop a new human-friendly active interface based on two key technologies: a mechanism for estimating human intentions to support natural communication, called the "Proactive Interface," and a tangible device using robot technology. We have developed a humanoid two-legged robot named "PICO-2" as a tangible telecommunication device for the proactive human interface. To achieve embodied telecommunication with PICO-2, we propose a new technique for tracking human gestures using a monocular video camera mounted on PICO-2, and natural gesture reproduction by PICO-2, which accounts for the differences in body structure between the user and the robot. Remote communication experiments were also carried out using two PICO-2 robots placed at distant campuses.

photo
PICO
PICO-2		 photo
PICO-2		 photo
PICO-2K
Mimic gestures ZMP balance control Mimic gestures Omnidirectional motion/td>
Motion tracking by monocular camera Remote communication experiments

Papers

  • Seiichi Uchida, Akihiro Mori, Ryo Kurazume, Rin-ichiro Taniguchi and Tsutomu Hasegawa,Logical DP Matching for Detecting Similar Subsequence, 8th Asian Conference on Computer Vision,November 2007.
  • Masato Nakajima, Seiichi Uchida, Akihiro Mori, Ryo Kurazume, Rin-ichiro Taniguchi, Tsutomu Hasegawa, and Hiroaki Sakoe, Motion Prediction Based on Eigen-Gestures, Proc. of the First Korea-Japan Joint Workshop on Pattern Recognition, pp.61-66, 2006.
  • Akiriho Mori, Seiichi Uchida, Ryo Kurazume, R.inichiro Taniguchi, Tsutomu Hasegawa, and Hiroaki Sakoe, Early Recognition and Prediction of Gestures toward Intelligent Man-Machine Interfaces, Proc. The Second Joint Workshop on Machine Perception and Robotics, CD-ROM, 2006.
  • Ryo Kurazume, Hiroaki Omasa, Seiichi Uchida, Rinichiro Taniguchi,Tsutomu Hasegawa, Embodied Proactive Human Interface ''PICO-2'', Proc. International Conference on Pattern Recognition, B04-0204, Aug 2006.
  • Akiriho Mori, Seiichi Uchida, Ryo Kurazume, R.inichiro Taniguchi, Tsutomu Hasegawa, and Hiroaki Sakoe, Early recognition and prediction of gestures, Proc. International Conference on Pattern Recognition, C02-0725, Aug 2006.
  • Yutaka Araki, Daisaku Arita, Rinichiro Taniguchi, Seiichi Uchida, Ryo Kurazume and Tsutomu Hasegawa, Construction of symbolic representation from human motion information, 10th Int. Conf. on Knowledge-Based & Intelligent Information & Engineering Systems (KES2006), 2006.
  • Rinichiro Taniguchi, Daisaku Arita, Seiichi Uchida, Ryo Kurazume, and Tsutomu Hasegawa, Human action sensing for proactive human interface: Computer vision approach ,Proceedings of International workshop on Processing Sensory Information for Proactive Systems (PSIPS 2004, Oulu, Finland)