Cheetah-Cub – a compliant quadruped robot

Cheetah-cub, a compliant quadruped robot

We are glad to present Cheetah-cub, a compliant quadruped robot with the size of a small house cat, or a young cheetah cubThe robot weights 1kg and is approximately 21cm long. It reaches 1.42m/s speed, almost seven body lengths per second. This makes Cheetah-cub robot the fastest running quadruped legged robot under 30kg.

Cheetah-Cub has several interesting features, especially when compared to larger and stiffer quadruped robot designs. 1) It is, to the best of our knowledge, the fastest of all quadruped robots below 30kg (in terms of Froude number and body lengths per second). 2) It shows self-stabilizing behavior over a large range of speeds with open loop control. 3) It is lightweight, compact, electrically powered. 4) It is cheap, easy to reproduce, robust, and safe to handle. This makes it an excellent tool for research of multi-segment legs in quadruped robots.

This work appears in the IJRR OnlineFirst June 2013 issue (Monday June 17, 2013). More information including pictures, videos, publications, and a more detailed technical description can be found below on this page. Videos can also be found on our epflbiorob Youtube channel. This work is funded by the European Commission through the AMARSI project.

Pictures

 

Copyright of pictures and videos: see the end of this page.

Research overview

This research describes the novel compliant quadruped robot Cheetah-cub, its implementation, and extensive testing and experimentation. This work appears in the IJRR OnlineFirst June 2013 issue. The paper includes topics such as Cheetah-cub’s bio-inspired leg and robot design, a locomotion controller based on coupled neuron-like oscillators (central pattern generator), extensive modeling of robot dynamics in an open dynamics engine based simulation environment (Webots), biomechanical experiments such as step-down perturbations and kinematic and dynamic gait measurements, and an extensive analysis of Cheetah-cub’s locomotion characteristics.

We show experimental results where Cheetah-cub robot reaches 1.42m/s speed, almost seven body lengths per second. This makes Cheetah-cub robot the fastest running quadruped legged robot under 30kg (normalized speed via Froude number FR=1.3=(speed^2)/(G times leg length) . The robot exhibits robust locomotion and self-stabilizing characteristics: even when it encounters perturbations at high speed such as a step-down, it will often successfully pass this obstacle (between 80% and 20% success rate depending on the height of the step down, with step-down heights up to 20% of the standing leg length). This robustness, and the results of further experiments, are indicators of the self-stabilizing properties of the robot’s leg and locomotion patterns: all experiments were conducted in an open-loop fashion.

Cheetah-cub robot is a 1kg machine of the size of a small house cat, or a young cheetah cub. The robot’s 15cm long legs were designed and implemented according to a blueprint suggested by researcher from Biomechanics (e.g. Witte et al., 2001). They observed that many quadrupedal, mammalian animals feature a distinguished functional three-segment front leg and hind leg design, and proposed a “pantograph” leg abstraction for robotic research. With Cheetah-cub, we pursue this blueprint, and designed, implemented, and tested two variations of such a pantograph leg design. Both implementations make use of in-parallel leg elasticity, through leg springs integrated in the pantograph and the robot’s feet. The passive response of mechanical springs in robot legs has shown to be a good approximation of the muscle-tendon complex in animal legs and humans. The robot’s legs were each actuated by two powerful RC servo motors, moving hip and knee joints.

Characteristics of Cheetah-cub robot
Maximum speed, vmax 1.42m/s (5.1km/h or 3.2 miles/h)
Froude number FR (v^2/G/lhip) 1.30
Body lengths per second 6.9
Gait type Trot
Active degrees of freedom 8
Mrobot 1100g (2.4 pounds)
Mactuators, sum 590g
lhip, standing height 0.158m (0.5 foot)
dshoulder-shoulder 0.1m
dhip-shoulder 0.205m
RC servo motor Kondo KRS2350 ICS (8x)
Stall torque RC servo 2Nm at 6V
Speed max RC servo 0.16s / 60deg at 6V
Control board RoBoard RB110
Power supply, tethered 8V to 14V

Work with Cheetah-cub is also a continuation of work at the Biorobotics laboratory researching central pattern generators (CPGs): networks of coupled neurons. CPGs are sophisticated circuits that can produce complex locomotor patterns while receiving only simple command signals from upper parts of the brain. Cheetah-cub is a natural follow-up to Amphibot and Salamandra robotica, which were  Biorob’s first CPG driven, lamprey-like and salamander-like robots. For Cheetah-cub robot we again implemented a CPG, for the generation of a variety of trot-gait like locomotion patterns. When designing Cheetah-cub’s CPG, we focused on biomechanically feasible control parameters: amplitude and offsets of hip oscillations, frequency of the locomotion cycle, and phase relationship between hip angle and knee angle. A very important parameter for animal locomotion is the leg cycle duty factor: it determines the amount of time a leg touches the ground and propels the animal forward, versus the full locomotion cycle time. For slower locomotion, the duty factor is typically higher, the leg stays longer on the ground than it swings through the air. With faster locomotion speed the duty factor becomes smaller. Duty factor is one control parameter in Cheetah-cub’s CPG controller.

The combination of Cheetah-cub robot’s bio-inspired leg design, and its CPG-based control allowed us to robustly test both components. We used an external motion capture system based on reflective markers to accurately measure robot position and orientation in space. Two force plates recorded ground reaction force data, and we recorded the power consumption of Cheetah-cubs actuation to calculate its electrical (metabolic) cost of transport.

 

Cheetah-cub robot has several interesting features, especially when compared to larger and stiffer quadruped robot designs. 1) It is, to the best of our knowledge, the fastest of all quadruped robots below 30kg (in terms of Froude number and body lengths per second). 2) It shows self-stabilizing behavior over a large range of speeds with open loop control. 3) It is lightweight, compact, electrically powered. 4) It is cheap, easy to reproduce, robust, and safe to handle. This makes it an excellent tool for research of multi-segment legs in quadruped robots.

 

Scientific publication (available online from Monday June 17th, 2013):

“Towards Dynamic Trot Gait Locomotion-Design, Control and Experiments with Cheetah-cub, a Compliant Quadruped Robot”. Sprowitz, Alexander; Tuleu, Alexandre; Vespignani, Massimo; Ajallooeian, Mostafa; Badri, Emilie; Ijspeert, Auke. The International Journal of Robotics Research 0278364913482017, first published June 17, 2013 as doi:10.1177/0278364913482017 . Draft version here.

See also:

Horse-Like Walking, Trotting, and Galloping derived from Kinematic Motion Primitives (kMPs) and their Application to Walk/Trot Transitions in a Compliant Quadruped Robot” by Federico L. Moro, Alexander Sprowitz, Alexandre Tuleu, Massimo Vespignani, Nikos G. Tsagarakis, Auke J. Ijspeert, and Darwin G. Caldwell, Biological Cybernetics. 2013 Jun;107(3):309-20.

Sample videos

The AVI files require the DivX codec see http://www.divx.com or the Xvid codec see http://www.xvid.org.

Cheetah-cub robot running in the hallway (realtime): (MPG, 4.4MB) (AVI xvid, 6.4MB) (HD MP4, 12.2MB)
Cheetah-cub different step-down runs (slowmo): (MPG, 7.0MB) (AVI xvid, 10.3MB) (HD MP4, 21MB)
Cheetah-cub high-speed run, leg designs (slowmo): (MPG, 6.8MB) (AVI xvid, 9.9MB) (HD MP4, 21.3MB)
Cheetah-cub robot running in the hallway (slowmo): (MPG, 10.1MB) (AVI xvid, 15.0MB) (MP4, 14.2MB)
Mechanical robustness, Cheetah-cub (realtime): (MPG, 1.6MB) (AVI xvid, 2.3MB) (HD MP4, 30.2MB)
Hallway-run with onboard camera (realtime): (MPG, 8.6MB) (AVI xvid, 7.1MB) (HD MP4, 110.7MB)

Publications

A. Spröwitz, M. Ajallooeian, A. Tuleu and A. Ijspeert. Kinematic primitives for walking and trotting gaits of a quadruped robot with compliant legs, in Frontiers in Computational Neuroscience, vol. 8, num. 27, p. 1-13, 2014.

A. Sproewitz, A. Tuleu, M. Vespignani, M. Ajallooeian and E. Badri et al. Towards Dynamic Trot Gait Locomotion—Design, Control and Experiments with Cheetah-cub, a Compliant Quadruped Robot, accepted in International Journal of Robotics Research.

F. L. Moro, A. Sprowitz, A. Tuleu, M. Vespignani and N. G. Tsagarakis et al. Horse-like walking, trotting, and galloping derived from kinematic Motion Primitives (kMPs) and their application to walk/trot transitions in a compliant quadruped robot, in Biological Cybernetics, p. 1–12, 2013.

M. Ajallooeian, A. Sproewitz, A. Tuleu and A. Ijspeert. Data-driven extraction of drive functions for legged locomotion: A study on Cheetah-cub robot. 6th International Conference on Adaptive Motion of Animals and Machines, 2013, Darmstadt, Germany, 2013.

M. Ajallooeian, S. Pouya, A. Sproewitz and A. Ijspeert. Central Pattern Generators Augmented with Virtual Model Control for Quadruped Rough Terrain Locomotion. 2013 IEEE International Conference on Robotics and Automation (ICRA 2013), Karlsruhe, Germany, 2013.

A. Tuleu, M. Ajallooeian, A. Spröwitz, P. Loepelmann and A. Ijspeert. Trot Gait Locomotion of A Cat Sized Quadruped Robot. International Workshop on Bio-inspired Robots, Nantes, France, 2011.

A. Sproewitz, M. Fremerey, K. Karakasiliotis, S. Rutishauser and L. Righetti et al. Compliant Leg Design for a Quadruped Robot. Dynamic Walking 2009, Vancouver, Canada, 2009.

S. Rutishauser, A. Sproewitz, L. Righetti and A. J. Ijspeert. Passive compliant quadruped robot using central pattern generators for locomotion control. International Conference on Biomedical Robotics and Biomechatronics, Scottsdale, 2008.

Past Cheetah student projects

  • Torque control of brushless motor for amphibious robotics, Maxim Bianchini (Semester project, 2019-2020)
  • Traversability pipeline for Krock 2, Thomas Havy (Semester project, 2019-2020)
  • Learning to move around combining reflexes and cpgs for a bio-inspired human model, Julien Couyoupetrou (Semester project, 2019-2020)
  • Design of an aeroponic system with plant growth monitoring for precision farming, Paul Callens (, 2019-2020)
  • Modulation of reflex parameters in human walking, Carla Nannini (Semester project, 2019-2020)
  • Modeling spasticity behavior in human pathological locomotion, Clara Viatte (Semester project, 2019-2020)
  • Building a 3D mouse treadmill for locomotion experiments, Maxime Riou (Semester project, 2019-2020)
  • MOUSE TREADMILL CONTROL, Didier Negretto (Semester project, 2019-2020)
  • Implementation of oscillator models and tools for analyzing coupled oscillators in FARMS, Savyaraj Deshmukh (Semester project, 2019-2020)
  • Design and improvement of the electronics for a hand exoskeleton, Yassine Ahaggach (Semester project, 2019-2020)
  • Firmware and Software Development for a Hand Exoskeleton, Beryl Yersin (Semester project, 2019-2020)
  • SEMG QUANTITATIVE ANALYSIS IN REHABILITATION PATIENTS USING MYO ARMBAND INTERFACE, Emma Bouton-Bessac (Semester project, 2019-2020)
  • SEMG QUANTITATIVE ANALYSIS IN REHABILITATION PATIENTS USING MYO ARMBAND INTERFACE #2, Tiago Cavaleiro Rodrigues (Semester project, 2019-2020)
  • Spinal Cord Maps of Spatiotemporal Alpha-Motoneuron Activation in Neuromechanical Simulations of Human Walking and Comparison wi, Marion Claudet (Internship, 2018-2019)
  • Networking and Data Collection for Amphibious Field Robotics via Robot Operating System (ROS), Adrien Chassignet (Semester project, 2018-2019)
  • Implementation of polynomial muscle joint coupling, Weipeng Li (Semester project, 2018-2019)
  • Sensitivity analysis on reflex parameters of a bio-inspired locomotion controller and mimetism based optimisation, Paul Prevel (Semester project, 2018-2019)
  • Studying the role of different muscle models on the low level control of lower limb, Oleg Gelfort (Semester project, 2018-2019)
  • Development of a passive tail test platform for amphibious robots, Simon Honigmann (Semester project, 2018-2019)
  • Multimodal controller design and characterization using somatosensory feedback for salamander locomotion, Blaise Etter (Semester project, 2018-2019)
  • Closed loop control of a flow tank, Yi-Shiun Wu (Semester project, 2018-2019)
  • Neck modeling to understand the mechanisms underlying posture control and balance, Corentin Puffay (Semester project, 2018-2019)
  • Interactive Console for Robot Control and Monitoring, Julien Lombardo (Semester project, 2018-2019)
  • Real-time of foot patch with RGB-D sensor and limited computational ressources, Emmanuel Pignat (Semester project, 2014-2015)
  • Design and integration of a multiaxis force moment sensor for a mobile quadruped platform, Nicolas Sommer (Semester project, 2010-2011)
  • Hardware Development on quadruped robots, Andreas Gassner (Semester project, 2010-2011)
  • Improvement of the Cheetah Locomotion Control, Alexandre Tuleu (Master (diploma) project, 2009-2010)
  • Segmented Leg Design in Robotics, Lorenz Küchler (Master (diploma) project, 2009-2010)
  • Cheetah II robot, Max Fremerey (Internship, 2008-2009)
  • Exploiting body dynamics on the Cheetah robot, Ivan Kviatkevitch (Semester project, 2008-2009)
  • Hardware for Cheetah robot, Emily Poplawski (Internship, 2008-2009)
  • Cheetah: compliant quadruped robot, Simon Rutishauser (Semester project, 2007-2008)
  • Development and Test of a Simulation Model for the Cheetah Robot, Martin Riess (Semester project, 2007-2008)

If you are interested in a semester or master student project at Biorob, please contact us! We are always interested in hearing from motivated students. A comprehensive list of past student projects can be found here, a list of open projects can be found here. We are also open for suggestions.

 

Copyright

Feel free to use pictures and movies without prior permission, provided they are credited as follows: Biorobotics Laboratory, EPFL. Please contact us by email at massimo.vespignani at epfl.ch or auke.ijspeert at epfl.ch if you would like additional information or higher resolution videos.