SNAKELIKE ROBOTS LOCOMOTIONS CONTROL R U S S I A N S T A T E S C I N T I F I C C E N T R E C E N T R A L R E S E A R C H A N D D E S I G N I N S T I T U T E O F R O B O T I C S A N D T E C H N I C A L C Y B E R N E T I C S SNAKELIKE ROBOTS LOCOMOTIONS CONTROL Oleg Shmakov CRDI RTC DEPARTMENT OF SPbSPU Joint Advanced Student School 2006 Saint Petersburg Course 5: Mechatronics – Foundations and Applications 5th APRIL 2006

Introduction Why snakelike robots? Biomechanics of snakes Review Mechanic model of snakelike robots Mathematical model of snakelike robots Hardware realization control Snakelike robot CRDI RTC Conclusion Nowadays snakelike robots are an actively upcoming area in robotics. Frequently the term snake like robot is applied to all hyper redundant robots; those are consisted from modules connected by active or passive joints. Among these devices the real snakelike robots are those that use the wave movements of the successive module chains for locomotion on hard surfaces and in liquids. At most attention of specialist's serpentine designs is explained by their desire to use the unique possibilities of the limblesses to adapt to surfaces and environments. These possibilities are relatively harmonize with the simple anatomy. The initial developments of the snakelike robots were executed in 70-s by Hirose's group of investigators. He made the analysis of limblesses motions experimental data and suggested mathematical description of the snake's instant form. The curve was called "serpenoid" and is used for the snakelike robot's control assignment. The first designs of Hirose's snakelike robots had modules with small passive wheels. The same design was used in projects of Gavin Miller. Robot that has no wheeled supports for its motion is more close to the biological analogs. The first successful developments of this type of robot appeared at the end of the 90-ies at Carnegie Mellon University. At the same time active development of algorithms of snakelike robot locomotion control was carried on. In the articles by Ostrovskiy, Chirikian, Choset, Dowling and other scientists various decisions for control generating that could provide some locomotion modes in the snakelike robots have been rendered. It is necessary to point out that for a long time a rational model of the snake's movement has not been presented. Such model would be very helpful to develop effective algorithms to control movements of the snakelike robots that have no wheels. One of quoted movement models has been offered in 1970. by A.I. Dobroluybov. The mechanical moving model of running waves of the deformation having local motionless contact, has allowed to give qualitatively true description of factors influencing on moving of a flexible body with use various locomotion modes. Unfortunately, it did not contain the proved mathematical description. In 2002-2004 A.A. Ivanov suggested some mathematical models of the flexible body dynamics and kinematics. These mathematical models allow to describe experimental data on snakes' kinematics within the limits of biological experiment precision. Having mathematical models let us make right constructive decisions not only on mechanical structure of the snakelike robot but also on the purposeful movement system control with the use of different modes of locomotion. I am going to take a review of the snakelike robot constructions as well as to report on biometrical requisites for development of such devices and to give a description of mechanical and mathematical models of the reptiles' movement. I would like to inform you about the real hardware of the snakelike robot movement control and to give a detailed description of the snakelike robot developed in our Scientific Research & Design Institute of Robotics and Technical Cybernetics. This model have already performed showing different modes of locomotion. Oleg A. Shmakov Snakelike robots locomotions control

Why snakelike robots? Advantages Stability Terrainability is the ability of a vehicle to traverse rough terrain Traction is the force that can be applied to propel a vehicle Redundant Simple anatomy Oleg A. Shmakov Snakelike robots locomotions control

Why snakelike robots? Applications Search and Rescue Examination blockages after earthquake Planet’s exploration Medical applications Examination hard-to-reach areas Tube inspection Bio Terrorist Remote sampling Military inspection …

Biomechanics of snakes Vertebrate Snake vertebral articulation is one of the most complex of all vertebrates. Real snakes have 100-400 vertebrae Oleg A. Shmakov Snakelike robots locomotions control

Biomechanics of snakes Locomotion modes Lateral undulation Sidewinding Propulsion (with creeping) Concertina Rectilinear movement Oleg A. Shmakov Snakelike robots locomotions control

Review Design snakelike robots Non-modular Modular WITH WHEELS WITHOUT WHEELS Oleg A. Shmakov Snakelike robots locomotions control

Review Hirose & Umetami - SERPENOID the waveform that the snake assumes during creeping movement is a curve which changes sinusoidally along the curvature of the body, and a formula for this, called a serpenoid curve. Oleg A. Shmakov Snakelike robots locomotions control

Review Hirose & Umetami – ACM III Active Cord Mechanisms – ACM Wheeled robots Robots that could perform lateral undulation Hirose’s development of modeling and control first derived expressions of force and power as functions of distance and torque along the curve described by the snake. Comparisons with natural snakes across constant friction surfaces showed close agreement between the serpenoid curve and the empirical data. Snakes quickly adapt locally to variations in terrain and environment. The control took the form of angle commands at each joint Oleg A. Shmakov Snakelike robots locomotions control

Review Hirose & Umetami – ACM R3 Oleg A. Shmakov Snakelike robots locomotions control

Review Carnegie Mellon University – Kevin Dowling Oleg A. Shmakov Snakelike robots locomotions control

Review Carnegie Mellon University – Biorobotics Lab Oleg A. Shmakov Snakelike robots locomotions control

Mechanic model of snakelike robots Dobroluybov A.I. «genetic relationship» wheels and waves and Snakes are using rolling motion ” Some points of a moving body or set of bodies during movement should vary periodically roles: mobile points become motionless and on the contrary. On character of this procedure of locomotion can be divided into two big classes: pacing when reference points of a body only during some moments of time pass from motionless in a mobile condition and back, and rolling when these transitions are carried out continuously. Snakes can move by pacing and rolling. Carry of points of a support of the essences, moving in the way rolling, can be various. Oleg A. Shmakov Snakelike robots locomotions control

Mechanic model of snakelike robots Ivanov A.A. Oleg A. Shmakov Snakelike robots locomotions control

Mechanic model of snakelike robots Ivanov A.A. Shank movement Lateral bending Rectilinear movement curving (without creeping) Concertina Propulsion (with creeping) Side winding Oleg A. Shmakov Snakelike robots locomotions control

Mathematical model of snakelike robots How to control? Random Search Hill climbing Simulated Annealing Neural Nets Response Surface Methods Genetic Algorithms Trigonometric forms Fourier Parametric curves Bayesian optimization algorithms DOWLING CONRO TANEV CMU – Biorobotics Lab Oleg A. Shmakov Snakelike robots locomotions control

Genetic algorithms for locomotions control Category Value Function set {sin, cos, +, -, *, /} Terminal set {time, segment_ID, Pi, random constant, ADF} Population size 200 individuals Selection Binary tournament, ratio 0.1 Elitism Best 4 individuals Mutation Random subtree mutation, ratio 0.01 Fitness Velocity of simulated Snakebot during the trial Trial interval 180 time steps, each time step account for 50ms of “real” time Termination criterion (Fitness >100) or (Generations>30) or (no improvement of fitness for 16 generations) Fitness convergence characteristics of 10 independent runs of GP for cases where fitness is measured as velocity in any direction (a) and snapshots of sample evolved best-ofrun sidewinding locomotion gaits of simulated Snakebot (b, c), viewed from above. The dark trailing circles depict the trajectory of the center of the mass of Snakebot. Timestamp interval between each of these circles is fixed and it is the same (10 time steps) for both snapshots. Trajectory of the central segment (cs) around the center of mass (cm) of Snakebot for a sample evolved best-of-run sidewinding locomotion (a) and traces of ground contacts (b). Oleg A. Shmakov Snakelike robots locomotions control

The determined approach of Ivanov A.A. Oleg A. Shmakov Snakelike robots locomotions control

Hardware realization control С0 С1 С2 18 19 16 17 14 15 12 13 10 11 8 9 6 7 4 5 2 3 0 1 RS232 Microcontrollers Oleg A. Shmakov Snakelike robots locomotions control

Hardware realization control microcontroller Control & power DC motors Feedback from sensing JOINT Power supply Sensors MAIN Microcontroller For all joints (CMU) Oleg A. Shmakov Snakelike robots locomotions control

Snakelike robot CRDI RTC Total mass 3 kg Length 1120 mm Width 65 mm Maximal course torque 0,3 Nm Maximal pitch torque 1,2 Nm Number of the links 16 Number of the joints 15 Number of the servos 30 Voltage 4,8 – 6 V Oleg A. Shmakov Snakelike robots locomotions control

System of snakelike robot control RS232 USB 1.1 Power supply Microcontrollers PC 4.8 - 6 volt Power supply servo 6 volt Camera Coding - MAX 232 “SnakeWheel -1” Oleg A. Shmakov Snakelike robots locomotions control

Structural control scheme 63 byte AA F1 1 CS 59 60 63 byte* Oleg A. Shmakov Snakelike robots locomotions control

Low level control fsend ≤ 50 Гц fbase = 1,5 Гц 60°/0,11 с T ≥ 2/3 1 channel 2 channel t, мс 10 мс 20 мс 0,9 1,5 2,1 fsend ≤ 50 Гц fsend = 30 – 60 Гц fbase = 1,5 Гц T ≥ 2/3 fsend ≤ 50 Гц 60°/0,11 с t φ T 20 – 40 main points Oleg A. Shmakov Snakelike robots locomotions control

Software scheme Entering parameters Forming Send Visualization Bloc Of Protection Forming Bloc Camera control Blok Changing movement Send RS-232 Oleg A. Shmakov Snakelike robots locomotions control

Software – snake-charmer 5 3 2 1 4 COURSE PITCH Oleg A. Shmakov Snakelike robots locomotions control

EXPERIMENT Lateral bending Amplitude corner by course 35 Amplitude corner by pitch 18 Quantity link which are using in course wave 8 4 Phase SPEED (max) (cm/sec) 2.5 SPEED (min) (cm/sec) 1 Oleg A. Shmakov Snakelike robots locomotions control

EXPERIMENT Lateral bending Oleg A. Shmakov Snakelike robots locomotions control

EXPERIMENT Side winding Amplitude corner by course 35 Amplitude corner by pitch 20 Quantity link which are using in course wave 8 Phase π/2 SPEED (max) (cm/sec) 4,3 SPEED (min) (cm/sec) 3 Oleg A. Shmakov Snakelike robots locomotions control

EXPERIMENT Side winding Oleg A. Shmakov Snakelike robots locomotions control

EXPERIMENT Motion on the given curve Oleg A. Shmakov Snakelike robots locomotions control

EXPERIMENT Motion on the given curve Oleg A. Shmakov Snakelike robots locomotions control

Conclusion Oleg A. Shmakov Snakelike robots locomotions control

? Thank you for your attention Oleg A. Shmakov Snakelike robots locomotions control