Walking Robots and the Central and Peripheral Control of Locomotion in Insects

From the abstract: "Control of insect walking can be considered hierarchical and modular." The interest of this quote, for me, will be determining what meaning he makes for each of those words. Modular means something like 'the system can be meaningfully conceptualized as having parts with well defined actions'. I think i can do better, his idea of module allows interactions between modules at the same scale in the heirarchy. It also implies that each module can specify parameters and parameter relationships that modules higher in the heirarchy are insulated from. 'heirarchical' means something like 'A module can either be a leaf or a node. Node modules influence the action of, and are influenced by, some set of other modules, said to be at a lower level, some of which may be leaves.' I have to look closer if this definition of module and heirarchy allows modules at the same level to interact. This definition of heirarchy allows the 'influenced by' because his thinking includes sensory feedback, so this definition of heirarchy does not imply strictly top-down control.

This paper is a sort of 'biology for engineers' , he even schematically abstracts out all the details of all the different ganglia by putting them into three functional boxes with arrows going back and forth. Delcomyn is an engineer's entomologist.

The lack of notes through most of this paper suggests that I fell asleep and haven't read most of it yet, but there are lots of notes at the end.

"Furthermore, it is also not clear which elements of insect locomotor control are actually critical for the extraordinary agility and adaptability exhibited by insects as they traverse rugged terrain.nevertheless, several points can be made:"

1. "insects exhibit hierarchical control of walking and use a modular organization of control elements during walking." Brain determines speed, and direction. but muscles and coordination of muscles is controlled 'locally' by "modules such as central pattern generators"
2. Sensors have much redundancy (and employ it robustly). Sensory input influences locomotion.
3. Neural circuits, controlling everything, are finely tuned and retuned, purposed and repurposed, all on the fly. This lends much versatility.
4. Biologically, there seems to be a spectrum between sensory (bottom up?) control and central (CPG) control. When a roach moves quickly, sensory input plays a minor role. When a stick insect moves slowly, CPG plays a very minor role and motion is determined largely by sensory input ("the insect requires sensory input at all times to generate a proper pattern of motor output.").Sensation is too slow to be efffective at high speeds of walking, but are essential at slow speeds. Earlier in the paper there are observations that a roach with a missing leg is less impaired at high speeds than at low speeds. Also, a roach with a sliver a wood replacing it missing leg is unimpaired at slow speeds (reflecting the role of sensory input at slow speeds) (this also reflects that I did read the whole paper, just did make marks).