<- main page 






February 11, 2007








  What it is (typically) A phyiscal entity that can move about in a purposeful way
  Autonomy Robots are not always autonomous, but the popular notion of robots is that they are completely autonomous
  Embodiment Robots are embodied de facto - typically when we say "robot" we mean "autonomous machine with a physical embodiment"
  Made from artificial materials Made by humans, as opposed to e.g. cyborgs, who are part human, part machine.
  The concept of "robot" Typcially is used in a very limited way; "cognitive robotics" is used to indicate "robots with more than just the ability to move".
  The word "robot" Comes from Karel Capek's play Rossum's Universal Robots; "robot" means "worker" Czech





















7.2 Robots: What Are They Good For?


  • Space exploration
  • Boring work in large areas (e.g. cleaning)
  • Dangerous areas (e.g. geological exploration, deep-sea diving)
  • Moving things around

Physical manipulation, esp. manufacturing:

  • sorting
  • assemby
  Application areas
  • Manufacturing
  • Telerobotics, telepresence, virtual reality (with varying degrees of autonomy)
  • Augmentation of human abilities
  • Transportation


















What are Robots Made of?
  • Complexity of arms counted in degrees of freedom (1-dimensional motion ability) - DoFs
  • Unusual to see robots with arms of larger than 6 DoFs, because search space becomes too large
  • Inverse kinematics: mathematics for computing joint angles based on a desired future location of manipulator
  • Hydraulic: Only for the largest, strongest (e.g. autonomous backhoe)
  • Pneumatics: requires pressurized air, containers heavy
  • Batteries: Major constraint on mobility and endurance
  • Computers small and efficient enough to do plenty of on-board processing
  • Use Wi-Fi for localized robots with more brainpower
  • Simplest to process and use: Sonars, Infrared sensors
  • Medium complexity: Laser range finders
  • Simple to use, very complex to process: Cameras, microphones
  • Touch sensors of various sorts - hard to use and process
  • Joint position sensing technologies - increasingly common in servos



















  Goal of navigation Find your way from A to B
  Constraints Obstacles prevent you from going in a straight line
  Planning Planning algorithms help you navigate around obstacles (planning will be covered later this week)
  • "Blind" navigation: Using knowledge of starting point, move according to an internal map
  • Common Problem: drift in sensors and position will accumulate errors in the position and orientation
  Landmark-based navigation Recognize textures and configurations of objects, walls, etc.
  Reactive navigation Use external environment to trigger next move(s)
  Map-based navigation

Use an internal map to determine how to get from A to B; examples:

  • 2D maps - very common. Problem: only represents the world at one level (e.g. 50 centimeters off the ground).
  • (Skundar's MultiMOT (multi-modal octree) representation of space)
  Obstacle representation
  • Bounding boxes around obstacles, to indicate areas safe for travelling
  • Gradient maps: the closer you are to an object the stronger its "repellent force"
  • Voronoi diagrams
  • Ideas from computer graphics (level of detail - LoD)
    • Helps reduce search space by ignoring details of the shape of things
  • Unreliable sensors (worse the further away objects are)
  • Integrate sensors with a-priori knowledge
  • Build up a usable map of terrain that is being navigated



















Controlling Robots: Robot Cognition

"Behavior-based AI"

Term used to describe "reactive" architectures with a tight coupling of sensors and actuators


Example typical topic

Autonomous vaccum cleaner (i.e. a caricature - but see iRobot's Roomba autonomous vacuum cleaner)

  Behavior-based AI: What it's good for
  • Robust simple behaviors in complex, realtime environments
  • Insect-like behaviors
  • Reactive planning
  • Instinctive behaviors
  • As a robust basic component in a larger system
  Subsumption control







iRobot Roomba autonomous vacuum cleaner, c.a. y 2002





















Example Robots
  • MIT
  • Main methodology: Subsumption architecture




  Attila / Genghis
  • MIT
  • Main methodology: Competence network; subsumption architecture






P1 - first Asimo prototype


ASIMO - c.a. 2003






TM-SUK female robot 2004



PINO - open robot platform



Robots can also swim

See also [robotuna]



Banryu robot guard "dog"




"Pet" robots are becoming increasingly popular [video]

Hobby robotics will take off in coming years. Example: Chronio [1] [2] [3] [4]

Missing in these robots is a sense of body stance, and they typically have little or no vision. Therefore, their movements are mostly scripted (dead-reconing) movements.


Humanoid torso powered by pneumatics


Fact or fiction? [video]