See the Snake Paper
The Advantages of Ribbed, Flexible Snake Robots
Abstract
Introduction
Snake Robotics Rationale
This stuff taken primarily from Dowling.
Obviously, there are a lot of potential use-cases for agile and robust snake robots:
- Stability
- "Terrainability"
- Traction
- Efficiency
- Size
- Redundancy
- Sealing
Disadvantages (should I mention these?)
- Payload
- Many Degrees of Freedom
- Thermal Control
- Speed
Applications
- Exploration
- Inspection
- Routing
- Medical
- Hazardous Environments
- Search & Rescue
- Reconnaissance
Overview of Prior Work
See Snake Robotics Research Index for the list
- Early 1970s Hirose
- ACM (2d)
- Joint actuation (I believe exclusively?)
- 1992 - Burdick and Chirikjian - joint actuation, path planning, etc
- 1995 - NEC "Quake Snake" aka Orochi
- Slim, joint actuation, active U-joint (novel form of a Hooke's joint [Dowling 27])
- "One of the best mechanical designs" - real applications, effective packaging, modularity of links
- Another version of the joint was used in a snake built for JPL
- 1995 - Nilsson rolling joint, right up a tree!
- 1996 - Paap GMD snake, with ribs and cords and everything!
- CMU Snakes
- UMich Snakes
- Gavin Miller's Snakes
Current Shortcomings
Motivate the paper. Rationale of snake robots. Lots of background on previous work. Skeleton and musculator, but we're going to treat the problem inside out (muscles first, then skeleton); most do it the other way, which is why we end up with U-joint robots. Problems with cost, reliability, robustness.
Actuation Strategies
Actuation at joints vs. actuation between ribs.
Joint Actuation Advantages
- Direct drive train (motors to joints)
- More accurate/precise position control with motors
- Easier to characterize
- Simple kinematics
- More familiar to roboticists
- Lots of prior work
Dowling says "Mature technologies for emulating characteristics of biological muscle lie in the future. However, electromagnetic motors offer many options, and the robot snake is designed around these technologies." (Dowling 4). However, they're also primarily designed around joint actuation as well. People have been working really hard to improve mechanism for more torque, power, etc. Especially Howie Choset over at CMU. But it's really hard. You're sort of locked into using motors and relatively complex geartrains.
Rib Actuation Advantages
- Supports a much broader array of actuator types (pneumatic, cables, linear actuators, "servo-brass-rod(??)" etc)
- Actuators can cross multiple ribs!
- Actuator attachment locations can be adjusted
- An explicit level-arm that can be larger, so smaller forces than gears/axels
- More biomimetic
- No complex levers/gears required
- Less precision required
- Less opportunity for actuator failure (reliability, cost, etc)
List of possible actuation techniques (motors in servos, ball screws, gears, cables, cross-tendon drive, pneumatics, hydraulics)
Structural Strategies
Snakes (with ribs and spinal cord) use a type of ball joint. But they have 100-500 ribs! Do we need to use discrete joints like this in the skeleton?
Also, rib actuation allows us to get rid of the U-joint concept altogether!
U-joints vs. flexible spine.
Flexible spine enables a few actuation strategies by being elastic, like cross-tendon drive; it provides a remedy for the hyper-redundant problem.
Flexible Spine Advantages:
- More natural look
- No sharp corners; snake doesn't get caught on stuff as easily
- Easier to add skin (continuous curvature)
- Can add non-actuated ribs for support of stuff, like skin
- Inherently elastic (can be advantage or disadvantage)
- Simpler
- Cheaper
- Harder to break
- Fewer Parts
- No lubrication required
- Inherently 3d
Flexible Spine Disadvantages:
- Buckling problem
- Elasticity un-controllable
Our Robots
Prototype 1: Lateral Undulation
Prototype 2: Sidewinder, rectilinear
Conclusion