Literature Review

A place for notes about books and articles which are relevant to work in the lab or general robotics background.

• de Santos, Garcia, Estremera. Quadrupedal Locomotion.
  • Stability - definition of static/dynamic stability and introduction of different stability margins defined by how far the projection of the CG is from any given edge of the support polygon. One of the better ones is the Energy Stability Margin, which is the minimum potential energy required to tumble the robot around the edges of the support polygon.
  • Gaits - Definition of continuous versus discontinuous gaits, including event sequence gates and singular qates (where two or more events can occur at the same time).
  • My Questions/Thoughts
    • The discussion of the control of quadrupeds when some areas of the terrain are considered forbidden reminded me of the possibility of not being able to use certain legs while traveling -- which then reminded me of three-legged dogs, which still manage dynamically-stable gaits without leg placement optimized for three legs. This is also related to that mine-activating crawler which is designed specifically to lose legs.
    • What dynamic effects limit a statically stable machine's performance?

• Frasca, Arena, Fortuna. Bio-Inspired Emergent Control of Locomotion Systems.
  • Introduction: Defines Central Pattern Generator as a hierarchical organization of the motor system that can produce a rhythmic motor pattern with no need for sensory feedback or descending control. Presents a model of control based on the leg motions of hexapods, specifically cockroaches and stick insects.

• Raibert, Brown. Experiments in Balance with a 2D One-Legged Hopping Machine.
  • Used control system to balance a one-legged machine in a 2D plane. Separated control problem into tree parts: vertical height, horizontal velocity, and angular attitude control.
  • Showed the response to hopping in place to a changing desired height, moving to a specific position and staying there despite external perturbation, and leaping over an object.
  • While the robot performed well on the first two, it was determined that the placement of the foot before the leap was vitally important to whether or not the leap was successful – and more research would need to be done. (reminds me of Big Dog video from Boston Dynamics and their leaping demo)
• Kaliyamoorthy, Quinn, Zill. Force Sensors in Hexapod Locomotion.
  • In unknown terrain, force control is more appropriate than position control (to keep the robot standing), so they apply analysis of cockroaches in walking and climbing to force sensors in a hexapod robot.
  • Force sensors keep track of four types of stimuli - amount and rate of change of force, direction during stance, sudden force decreases (indicating slipping), and feedback during support or propulsion.

-- AnnaShedletsky? - 10 Aug 2007