Robotics

Reprise: piano movers’ problem. Locomotion

Lab session

Homework: postings. Work on projects.

Complexity

• Study of how long problem can take

• Example: certain sorts may take up to square of N steps, for input size N; others take less: N*log2 (N).

• Binary search (dividing the search space) means that any search limited by log2(N) steps.

NP complete problems• There is a set of problems for which there

hasn’t been found a good solution.– Where good means doing in polynomial time.

• Solutions for these problems can be checked in polynomial time.

• Examples: graph coloring, traveling salesman, logical satisfiability

• If one of these problems can be done, they all can be done.

The piano movers’ problem

• Moving something, such as a robot with payload (what it needs to carry) through a space with obstacles.

• Aka Motion planning• General problem is at least as hard as NP

complete problems and may be harder!• One approach is to grow the obstacles

while shrinking the robot and then go from corner to corner

Locomotion

• What is the problem• What is requirement for stability

– Static stability: capacity to return to standing if moved slightly

– Dynamic stability: capacity while moving to return to (steady, desired) movement when pushed

• What (where) is the control– What are the levels of control

May have high-level goal supplied and then device achieves that goal through local action

Distinct problems/challenges:Navigate over a particular terrain

• Level, smooth floors inside building– robot competitions– Manufacturing– Homes, hospitals, nursing homes, etc.

• Streets– light payloads– Heavy/heavier payload: autonomous cars

• In and out of streets & buildings– Rescue, detecting and disarming bombs, …

• Ground (including moon, Mars, etc.)

More challenges

• Combine travel over floor with climbing stairs in buildings (and other tasks such as climbing walls?)

• Imitate human locomotion – to replace [human] stuntman in movies– for testing of shoes and other gear

• More ???

Different problem: Prosthetic limbs

• Varying need to resemble somewhat what is being replaced vs

• Specialization for task– Sports limbs:

Already controversy: double leg replacementallowed for longer legs

http://www.spectrum.ieee.org/jun05/1217

Oscar Pistorius

Skiing

• Trak III:– Remove leg and use outriggers– Critical issue is how to remove, protect stump

and then re-attach limb

Skiing

• Sit ski

Artificial limbs issues

• Physical connection to body– Permanent?– Temporary but want long duration

• Control connection to nervous system for control– Same– Other

• Sensory input for performance of task

Hand

Goal is for hand to be able to grasp without crushing object or letting it slip

http://www.newscientisttech.com/article/dn10785-robotic-hand-has-a-builtin-slip-sense.html

Serpentine

• Problems– Surgery– Disarm bombs, mines

• Land or water

– Rough terrain

http://www.primidi.com/2005/03/23.html

Note

• Many of the experimental robots still are controlled teleoperator fashion

• Though…some have complex distributed control systems

Mobile robot (Ralph Hollis)

http://www.msl.ri.cmu.edu/projects/

General goal: mobile, dynamically stabileagile to bepersonal assistant

Robots in Scansorial Environment

(also Hollis lab) 6 legged climbing machine

Haptic devices

Touch

• contact

• force

• Note: may always be contact, for example, air, breeze. May need to quantify to distinguish.

Lab & Homework

• Continue to work on building project

• Postings