Gait Pattern Generation of Hexapod-Type Microrobot Using Interstitial Cell Model Based Hardware Neural Networks IC

Kurosawa, Mika; Sasaki, Takahiro; Ohara, Masaya; Tanaka, Taisuke; Yuki, Hideaki; Kaneko, Makoto; Uchikoba, Fumio; Saeki, Katsutoshi; Saito, Ken

doi:10.23919/icep.2019.8733603

Cited by 5 publications

(2 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Heuristics allow for blind-walking, i.e., the robot is able to walk through irregular terrain based only on proprioceptive information [26]. Neural networks can be implemented in hardware with miniaturization and energy efficiency as the main objective [27]. The Virtual Model Control [27] is one of the most relevant heuristics in order to obtain a model from experimental data, simulating the same dynamic behavior of complex mechanical systems using much simpler components such as springs, dampers, masses, etc.…”

Section: Heuristic Methodsmentioning

confidence: 99%

“…Neural networks can be implemented in hardware with miniaturization and energy efficiency as the main objective [27]. The Virtual Model Control [27] is one of the most relevant heuristics in order to obtain a model from experimental data, simulating the same dynamic behavior of complex mechanical systems using much simpler components such as springs, dampers, masses, etc. The resulting model is less complex but accurate enough to compute the forces acting on the robot body [28,29].…”

Section: Heuristic Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Kinematic Tripod (K3P): A New Kinematic Algorithm for Gait Pattern Generation

Soto-Guerrero,

Ramírez-Torres,

Rodriguez-Tello

2024

Applied Sciences

View full text Add to dashboard Cite

Insects are good examples of ground locomotion because they can adapt their gait pattern to propel them in any direction, over uneven terrain, in a stable manner. Nevertheless, replicating such locomotion skills to a legged robot is not a straightforward task. Different approaches have been proposed to synthesize the gait patterns for these robots; each approach exhibits different restrictions, advantages, and priorities. For the purpose of this document, we have classified gait pattern generators for multi-legged robots into three categories: precomputed, heuristic, and bio-inspired approaches. Precomputed approaches rely on a set of precalculated motion patterns obtained from geometric and/or kinematic models that are performed repeatedly whenever necessary and that cannot be modified on-the-fly to adapt to the terrain changes. On the other hand, heuristic and bio-inspired approaches offer on-line adaptability, but parameter-tuning and heading control can be difficult. In this document, we present the K3P algorithm, a real-time kinematic gait pattern generator conceived to command a legged robot. In contrast to other approaches, K3P enables the robot to adapt its gait to follow an arbitrary trajectory, at an arbitrary speed, over uneven terrain. No precomputed motions for the legs are required; instead, K3P modifies the motion of all mechanical joints to propel the body of the robot in the desired direction, maintaining a tripod stability at all times. In this paper, all the specific details of the aforementioned algorithm are presented, as well as different simulation results that validate its characteristics.

show abstract

Section: Heuristic Methodsmentioning

confidence: 99%