This document describes an experiment on a quadrapedal robot designed for self-healing capabilities. The robot consists of a body with four legs, each with two joints, and is equipped with sensors and actuated by servomotors. The robot uses an internal neural network model to autonomously synthesize predictive models of its own structure. The experiment involves introducing damage scenarios to test the robot's ability to recover through self-modeling, exploratory action, and adapting its target behavior based on its new configuration. The robot is able to detect changes to its morphology and continually update its internal model through exploration and hypothesis testing to compensate for damage.

1. By: Khemant Kumar Naik
Roll No: 08635
Guide: Banchha Nidhi Das

2. INTRODUCTION
• Self healing robots have ability to heal minor injuries.
• A robot is able to study its own structure through self
directed exploration.
• The concept used is multiple competing internal models and
generation of actions

3. THE STARFISH ROBOT-CHARACTERIZING
THE TARGET SYSTEM
 The target system is a quadrapedal,
articulated robot
 It consist of a rectangular body and
four legs attached to it with a hinge
joint
 Each leg has a upper and lower leg
attached together with a hinge joint

4. • All eight hinge joints are actuated with high torque
servomotors
• Servo drives capable of producing max 200 ounceinches
of torque and 60o per second speed
• Robot can automatically synthesize predictive models of
its own topology
• Equipped with suite of different sensors

5. SELF MODELLING CHALLENGES
• Robots depend on internal maps and sensory data to
update their location
• In a changing environment the robot has to navigate with
uncertainty
• Situation becomes even worse if the robot’s own shape
and configuration can change

6. PHASES IN SELF HEALING
 Self –Model synthesis
 Exploratory Action synthesis
 Target Behaviour synthesis

9. STEPS INVOLVED….
 Exploration phase: Controller evolution
 Physical robot failure
 Estimation Phase: Damage hypothesis evolution

10. EXPERIMENTAL SETUP
 The quadrapedal robot has eight degrees of freedom
 Two one degree of freedom rotational joints per leg:
one at the shoulder and one at knee
 It has 4 binary touch sensors and also 4 angle sensors
 All joints are actuated by a torsional motor

12. THE CONTROLLERS
 Robots are controlled by a
neural network
 There are 3 layers in the
neural network-input
layer,hidden layer,output
layer

13. DAMAGE SCENARIOS

14. DAMAGE RECOVERIES

15. Disadvantages …
 Creation of working model is costly
 Designing is complex and time taking
 Yet a perfect industrial model is not produced

16. Conclusion
 Although the possibility of autonomous self-modeling
has been suggested, it has much to perfect
 The robot's abilities suggest a similarity to human
thinking as the robot tries out various actions to figure
out the shape of its world.
 This suggests that future machines may be able to
continually detect changes in their own morphology or
structure

17. Reference …
 www.jyoti.debashish.info
 www.scribd.com
 http://en.wikipedia.org/wiki/Self-Reconfiguring_
Modular_Robotics
 http://ccsl.mae.cornell.edu/research/selfmodels/

18. Thank You
Any Queries ?