Seminar

1
Seminar

• Reading- Intuitive Control of a Planar Biped
  Walking Robot
• Jerry Pratt and Gill Pratt

2
Introduction

• How is walking stabilised in Intuitive Control
  ?
• What is Virtual Model Control ?
• What is the virtual toe point ?

3
Intuitive Control

• What is the main criticism of intuitive
  controllers
• How can automatic learning techniques be used for
  intuitive controllers and what are the benefits
  of doing so ?

4
Control Strategies

• Discuss how height is stabilised in a biped.
• Discuss how pitch is stabilised in a biped.
• Discuss how speed is stabilised in a biped.
• Discuss how the swing leg placement is
  controlled in a biped.
• Discuss how the the support leg transitions are
  controlled in a biped

5
Biped Kinematics

• What does it mean to say that the virtual toe
  point is a commanded quantity, not a measured
  one ?
• Derive the forward kinematic map from frame A
  to frame B
• Derive the Jacobian ABJ.

6
Implementing Virtual Actuators

• What will be the joint torques (?t, ?a, ?k, ?h)
  required to produce a force of (fx, fz, f?) on
  the robot at point B ?
• Given that the joint torque ?t is zero derive a
  constraint equation for (fx, fz, f?).
• Derive the relationship between the remaining
  joint torques (?a, ?k, ?h) and the virtual force
  applied to B for (fz, f?).

7
Biped Control

• The calculation of the virtual force does not
  include any terms for the robot mass and angular
  moments of inertia. How would you stop the robot
  falling ?

8
Dual Leg Implementation

• How can the joint torque problem be solved in the
  double support phase ?
• Are there any ways that this control can be
  improved ?

9
Spring Flamingo Implementation

• What constrains the spring flamingo robot to be a
  2D biped ?
• How do the series elastic actuators work ?
• What are the advantages and disadvantages of
  having the motors in the body and using cable
  drives to power the joints ?

10
Experimental Results

• Discuss the robustness properties of the spring
  flamingo robot and the walking control algorithm
  ?
• How would the algorithms be extended for a 3D
  biped robot system ?

11
Simulation Constructions Set

• Examine the Spring Flamingo and m2 simulations
  using the Yobotics Simulation Construction Set.
  (evaluation version available from
  www.Yobotics.com)
• Discuss the merits and problems of developing
  controllers in simulation and then transferring
  them to real robot systems ?