Key points: Why robots need self-sensing Sensors for proprioception in biological systems

1. Lecture 16: (20/11/09) Sensing self-motion Key points: • Why robots need self-sensing • Sensors for proprioception • in biological systems • in robot systems • Position sensing • Velocity and acceleration sensing • Force sensing • Vision based proprioception Michael Herrmann michael.herrmann@ed.ac.uk, phone: 0131 6 517177, Informatics Forum 1.42

2. Why robots need self-sensing • For a robot to act successfully in the real world it needs to be able to perceive the world and its relation to the world. • The state of the robot is not entirely up to the robot itself, but also reflects external events. Thus, information about the “body” is an important source of information about the world • Another use of proprioceptive information is stabilization and smoothing of planned movements against perturbations • In particular, to control its own actions, it needs information about the position and movement of its body and parts. • Our body contains at least as many sensors for our own movement as it does for signals from the world.

3. Proprioception: Detecting our own movements • To control our limbs we need feedback:Kinesthesia • Muscle spindles • where: length • how fast: rate of stretch • Golgi tendon organ • how hard: force

5. Proprioception: Detecting our own movements • To control our limbs we need feedback on where they are. • Muscle spindles • Golgi tendon organ • Pressure sensors in skin Pacinian corpuscle – transient pressure response

6. Proprioception (cont.) • To detect the motion of our whole body have vestibular system based on statocyst • Statolith (calcium nodule) affected by gravity (or inertia during motion) causes deflection of hair cells that activate neurons

7. Describing movement of body Requires: • Three translation components • Three rotatory components

8. Vestibular System Utricle and Saccule detect linear acceleration. Semicircular canals detect rotary acceleration in three orthogonal axes Fast vestibular-ocular reflex for eye stabilisation

9. Robert J. Peterka (2009) Comparison of human and humanoid robot control of upright stance. Journal of Physiology – Paris103, 149–158

10. Using proprioceptive information Control Proprioception Exteroception Efference copy body surface

11. For a robot: Need to sense motor/joint positions with e.g.: • Potentiometer (current measures position) • Optical encoder (counts axis turning)‏ • Servo motor (with position feedback)‏

12. For a robot: • Velocity by position change over time or other direct measurement: Tachometer • E.g. using principal of dc motor in reverse: voltage output proportional to rotation speed • (Why not use input to estimate output…?)‏ • Acceleration: could use velocity over time, but more commonly, sense movement or force created when known mass accelerates, i.e. similar to statocyst

13. Gyroscope:uses conservation of angular momentum Accelerometer:measures displacement of weight due to inertia There are many alternative forms of these devices, allowing high accuracy and miniaturisation (e.g. ceramic piezo gyros)

14. Inertial Navigation System (INS) • Three accelerometers for linear axes • Three gyroscopes for rotational axes (or to stabilise platform for accelerometers) • By integrating over time can track exact spatial position • Viable in real time with fast computers • But potential for cumulative error

15. For a robot: To sense force: e.g. • Strain gauge – resistance change with deformation • Piezoelectric – charge created by deformation of quartz crystal (n.b. this is transient)

16. For a robot: • Various other sensors may be used to measure the robot’s position and movement, e.g.: • Tilt sensors • Compass • GPS • May use external measures e.g. camera tracking of limb or robot position

17. Some issues for sensors • What range, resolution and accuracy are required? How easy to calibrate? • What speed (i.e. what delay is acceptable) and what frequency of sampling? • How many sensors? Positioned where? • Is information used locally or centrally? • Does it need to be combined?

18. Haptic perception – combines muscle & touch sense

19. Vision as proprioception? • An important function of vision is direct control of motor actions • Test: standing on one leg with eyes closed or standing up ...

20. The ‘swinging room’ - Lee and Lishman (1975)

21. Optical flow

22. Optical flow:Heading = focus of expansion …provided that it can discount flow caused by eye movements

23. Optical flow: Flow on retina = forward translation + eye rotation Flow-fields if looking atx while moving towards+ Bruce et al (op. cit) Fig 13.6

25. P From optical flow to time to contact P = distance of image from centre of flow X = distance of object from eye V = velocity of approach Y = velocity of P on retina t = P/Y = X/V rate of image expansion = time to contact Lee (1980) suggested visual system can detect tdirectly and use to avoid collisions e.g. correct braking.

26. Using expansion as a cue to avoid collision is a common principle in animals, and has been used on robots • E.g. robot controller based on neural processing in locust – Blanchard et. al. (2000)‏

27. Proprioceptive control

28. Proprioceptive control

29. Summary • Have discussed a variety of natural and artificial sensors for self motion • Have hardly discussed how the transduced signal should be processed to use in control for a task. • E.g. knowing about muscle and touch sensors doesn’t explain how to manipulate objects

30. Dimensions of robotics • Defining goals: Tasks or models • Reaching goals: programming or learning • Reason or emotions • Evaluation of performance • Energy consumption • Social issues: • Dynamical systems for control • Design principles

31. 1. Biorobotics • Robots as models of animal behaviour • Proof of (functional) principle • Bio-inspired robotics • Biomorphic engineering • Service robots • Prosthetics • Human-robot interaction

32. 2. Programming vs. Learning O. Lebeltel, 1996

33. 2.Programming vs. Learning O. Lebeltel, 1996

34. Programming and Learning for control • Action languages (R. Reiter) • Middleware concepts • Machine learning algorithms • Objective functions • Self-organisation of behaviour • Evolution and development • Reinforcement learning • Neural networks • Artificial emotions, consciousness Methods from Comp. Sc. Engineering Math Physics Biology Psychology

35. The uncanny valley(Masahiro Mori, 1970)‏ Repliee Q1 and Geminoid (H. Ishiguro, U Osaka, 2005, 2007)

36. 3. Emotion vs. Reason Emotions for robots: • Interaction with humans • Internal evaluation • Centralised supervision • Kansei (emotion) engineering Reason for robots: cf. 2. and previous lectures

37. 4. Performance: Competition vs. Measurement • DARPA Grand Challenge • RoboCup: Robot Soccer & Rescue • Climbing, underwater, fire fighting, ... • RunBot: Fastest robot on two legs • Service limits, running costs, monitoring and support, flexibility, upgradability

38. 5. Energy consumption • Super-human efficiency in certain tasks • Inspiration from biology: Passive dynamics in walking, energy re-use by springs, locking mechanisms for posture maintenance, modularity, hibernation • Development of enduring batteries • Alternative energies: Solar robots • Fly-eating robot (UWE, 2004)

39. 6. Social robots • Division of labour, specialised hardware • Communication, cooperation, collaboration • Collaboration gain (super-linear increase with number of robots?) • Understanding language and social behavior • Swarms intelligence from many very simple robots • Human-Robot workgrounps

40. 7. Dynamical systems vs. control • Closed perception-action loop • Everything is in the senses • Evolution • No planning, no representation • Exploratory • Potentially interesting Feed-forward, feed-back Objective-driven, uses prior knowledge Design Planning reqired for complex goals Dependability Potentially useful

41. 8. Distributed vs. centralized • Modularity on all levels • Re-configurability • Fast local computations • Communication partially replaced by local decisions • Bio-inspired solutions Monitoring Simplicity Debugging Communication less demanding

42. 9. Areas of applications • Assembly, manufacturing, manipulation • Remote operation, exploration, rescue • Science and education • Prosthetics, orthotics, surgery, therapy • Service, transport, surveillance • Entertainment, toys, sports • Military

43. More dimensions • Vision • Sensing and Acting • Locomotion, reaching and grasping • Dynamics and kinematics • Control • Internal organization, architectures