Towards Agility: Definition, Benchmark and Design Considerations for Small, Quadrupedal Robots

Agile quadrupedal locomotion in animals and robots is yet to be fully understood, quantified or achieved. An intuitive notion of agility exists, but neither a concise definition nor a common benchmark can be found. Further, it is unclear, what minimal level of mechatronic complexity is needed for this particular aspect of locomotion. In this thesis we address and partially answer two primary questions: (Q1) What is agile legged locomotion (agility) and how can wemeasure it? (Q2) How can wemake agile legged locomotion with a robot a reality? To answer our first question, we define agility for robot and animal alike, building a common ground for this particular component of locomotion and introduce quantitative measures to enhance robot evaluation and comparison. The definition is based on and inspired by features of agility observed in nature, sports, and suggested in robotics related publications. Using the results of this observational and literature review, we build a novel and extendable benchmark of thirteen different tasks that implement our vision of quantitatively classifying agility. All scores are calculated from simple measures, such as time, distance, angles and characteristic geometric values for robot scaling. We normalize all unit-less scores to reach comparability between different systems. An initial implementation with available robots and real agility-dogs as baseline finalize our effort of answering the first question. Bio-inspired designs introducing and benefiting from morphological aspects present in nature allowed the generation of fast, robust and energy efficient locomotion. We use engineering tools and interdisciplinary knowledge transferred from biology to build low-cost robots able to achieve a certain level of agility and as a result of this addressing our second question. This iterative process led to a series of robots from Lynx over Cheetah-Cub-S, Cheetah-Cub-AL, and Oncilla to Serval, a compliant robot with actuated spine, high range of motion in all joints. Serval presents a high level of mobility at medium speeds. With many successfully implemented skills, using a basic kinematics-duplication from dogs (copying the foot-trajectories of real animals and replaying themotion on the robot using a mathematical interpretation), we found strengths to emphasize, weaknesses to correct and made Serval ready for future attempts to achieve even more agile locomotion. We calculated Servalâs agility scores with the result of it performing better than any of its predecessors. Our small, safe and low-cost robot is able to execute up to 6 agility tasks out of 13 with the potential to reachmore after extended development. Concluding, we like to mention that Serval is able to cope with step-downs, smooth, bumpy terrain and falling orthogonally to the ground.

Towards rich motion skills with the lightweight quadruped robot Serval

Peter Eckert, Tomislav Horvat, Auke Ijspeert, Anja Eva Maria Schmerbauch

Bio-inspired robotic designs introducing and benefiting from morphological aspects present in animals allowed the generation of fast, robust, and energy-efficient locomotion. We used engineering tools and interdisciplinary knowledge transferred from biology to build low-cost robots, able to achieve a certain level of versatility. Serval, a compliant quadruped robot with actuated spine and high range of motion in all joints, was developed to address the question of what mechatronic complexity is needed to achieve rich motion skills. In our experiments, the robot presented a high level of versatility (number of skills) at medium speed, with a minimal control effort and, in this article, no usage of its spine. Implementing a basic kinematics-duplication from dogs, we found strengths to emphasize, weaknesses to correct, and made Serval ready for future attempts to achieve more agile locomotion. In particular, we investigated the following skills: walk, trot, gallop, bound (crouched), sidestep, turn with a radius, ascend slopes including flat ground transition, perform single and double step-downs, fall, trot over bumpy terrain, lie/sit down, and stand up.

SAGE PUBLICATIONS LTD2020

Control of Bio-Inspired Sprawling Posture Quadruped Robots with an Actuated Spine

Tomislav Horvat

Sprawling posture robots are characterized by upper limb segments protruding horizontally from the body, resulting in lower body height and wider support on the ground. Combined with an actuated segmented spine and tail, such morphology resembles that of salamanders or crocodiles. Although bio-inspired salamander-like robots with simple rotational limbs have been created, not much research has been done on kinematically redundant bio-mimetic robots that can closely replicate kinematics of sprawling animal gaits. Being bio-mimetic could allow a robot to have some of the locomotion skills observed in those animals, expanding its potential applications in challenging scenarios. At the same time, the robot could be used to answer questions about the animal's locomotion. This thesis is focused on developing locomotion controllers for such robots. Due to their high number of degrees of freedom (DoF), the control is based on solving the limb and spine inverse kinematics to properly coordinate different body parts. It is demonstrated how active use of a spine improves the robot's walking and turning performance. Further performance improvement across a variety of gaits is achieved by using model predictive control (MPC) methods to dictate the motion of the robot's center of mass (CoM). The locomotion controller is reused on an another robot (OroBOT) with similar morphology, designed to mimic the kinematics of a fossil belonging to Orobates, an extinct early tetrapod. Being capable of generating different gaits and quantitatively measuring their characteristics, OroBOT was used to find the most probable way the animal moved. This is useful because understanding locomotion of extinct vertebrates helps to conceptualize major transitions in their evolution. To tackle field applications, e.g. in disaster response missions, a new generation of field-oriented sprawling posture robots was built. The robustness of their initial crocodile-inspired design was tested in the animal's natural habitat (Uganda, Africa) and subsequently enhanced with additional sensors, cameras and computer. The improvements to the software framework involved a smartphone user interface visualizing the robot's state and camera feed to improve the ease of use for the operator. Using force sensors, the locomotion controller is expanded with a set of reflex control modules. It is demonstrated how these modules improve the robot's performance on rough and unstructured terrain. The robot's design and its low profile allow it to traverse low passages. To also tackle narrow passages like pipes, an unconventional crawling gait is explored. While using it, the robot lies on the ground and pushes against the pipe walls to move the body. To achieve such a task, several new control and estimation modules were developed. By exploring these problems, this thesis illustrates fruitful interactions that can take place between robotics, biology and paleontology.

EPFL2018

Legged locomotion with spinal undulations

Konstantinos Karakasiliotis

Legged locomotion with spinal undulations is a topic that has not received much attention yet in robotics, though, vertebrates depend on both their limbs and spine to move efficiently in their environments. In this thesis, my aim is to get more insights into the role and function of these two body parts as well as their coordination. Biology provides several answers to these questions. Salamanders and lizards are great biological paradigms for studying animal locomotion with limbs and flexible spine. Not only do they have a highly flexible spine and limbs that can sometimes vary in strength or even number, but they also demonstrate a rich repertoire of locomotor modes. They can walk on land, while they can swim and walk in water. This provides many more examples on the ways that limbs and spine function and coordinate. Biorobotics is a growing field of robotics that aims at answering questions that cannot be answered from animal observations. Those questions often deal with the development of a specific function of a body part. In my studies, I initially used a bio-inspired amphibious salamander-like robot Salamandra robotica II to extract basic principles of salamander locomotion. Although the improvements of the limb and tail designs enhanced the robot’s performance to a level that was surprisingly close to the salamander’s, several limitations were identified owing to the simplicity of the bio-inspired design. These limitations motivated me to design a new amphibious salamander-like robot, Pleurobot, which is capable of replicating the movements of salamanders with reasonable accuracy. A novel methodology was also used during this process which proved to be effective but simple at the same time. Three-dimensional kinematics of the salamander’s skeleton were recorded and used to guide the design process and optimization of the robot’s morphology and joints’ topology. The raw angular kinematics, extracted from X-ray biplanar videos of salamanders, were then used to drive the joints of Pleurobot, resulting in natural-looking and effective locomotion. The main contributions of this thesis are two-fold: i) biological data and robotic prototypes that improve our understanding on sprawling locomotion and ii) a successful methodology for robotic design for more animal-like robots and agile control