The reconstruction of mechanisms behind odour-tracking behaviours of animals is expected to enable the development of biomimetic robots capable of adaptive behaviour and effectively locating odour sources. However, because the behavioural mechanisms of animals have not been extensively studied, their behavioural capabilities cannot be verified. In this study, we have employed a mobile robot driven by a genuine insect (insect-controlled robot) to evaluate the behavioural capabilities of a biological system implemented in an artificial system. We used a male silkmoth as the 'driver' and investigated its behavioural capabilities to imposed perturbations during odour tracking. When we manipulated the robot to induce the turning bias, it located the odour source by compensatory turning of the on-board moth. Shifting of the orientation paths to the odour plume boundaries and decreased orientation ability caused by covering the visual field suggested that the moth steered with bilateral olfaction and vision to overcome the bias. An evaluation of the time delays of the moth and robot movements suggested an acceptable range for sensory-motor processing when the insect system was directly applied to artificial systems. Further evaluations of the insect-controlled robot will provide a 'blueprint' for biomimetic robots and strongly promote the field of biomimetics.

SUMMARY: By combining optical triangulation with the comb-fringe technique and dual-channel telemetry, wing kinematics and body attitudes accompanying muscle activities of free-flying male hawkmoths were recorded synchronously when they performed flight manoeuvres elicited by a female sex pheromone. The results indicate that the wing leading edge angular position at the ventral stroke reversal, which can be decomposed by two orthogonal angular parameters (a flapping angle and a deviation angle), is well controllable. Two specific flight muscles, the dorsal-ventral muscle (DVM, indirect muscle, a wing elevator) and the third axillary muscle (3AXM, direct muscle, a wing retractor), can modulate the flapping angle and the deviation angle,respectively, by means of regulating the firing timing of muscle activities. The firing timing can be expressed by the firing latency absolutely, which is just before the timing of ventral stroke reversal. The results illustrate that lengthening the firing latency of the DVM and of the 3AXM can increase the flapping angle and the deviation angle, respectively, which both strengthen the downstroke at the ventral stroke reversal. The relationship of bilateral asymmetry shows that the bilateral differences in the firing latency of the DVM and of the 3AXM will cause bilateral differences in the wing position,which accompany the variations of yaw and roll angles in time course. This implies the contribution of the two muscles to active steering controls during turning or banking, though the DVM being an indirect muscle was generally treated as a power generator. Finally, the relationship between the pitch angle and the 3AXM latency, deduced from the relationships between the pitch angle and the deviation angle and between the deviation angle and the 3AXM latency, shows that lengthening the 3AXM latency can increase the pitch angle at the ventral stroke reversal by moving the wing tip far away from the centre of gravity of the body, which indicates a functional role of the 3AXM in active pitching control.

In this study, we propose a robust chemical plume tracing (CPT) algorithm for various environments. We developed this algorithm based on the behavior of insects that have CPT abilities globally. However, it is difficult to accurately estimate the behavior pattern from trajectory data because the trajectory contains a high level of measurement noise. We employed flight muscle electromyograms, whose motor commands were assumed to be generated in the same ganglion as the behavior, to estimate the behavior pattern using a support vector machine. By using the estimated results, we modeled the time-varying behavioral change of insects and verified the effectiveness of the phenomenon for CPT using the constructive approach. We named this time-varying CPT algorithm as time-varying moth-inspired algorithm. From the CPT experiment results using the robot, we found that the localization success rate improved by changing the behavior in a time-variant manner.

The interaction between neuromuscular systems and body mechanics plays an important role in the production of coordinated movements in animals. Lepidopteran insects move their wings by distortion of the thorax structure via the indirect flight muscles (IFMs), which are activated by neural signals at every stroke. However, how the action of these muscles affects thorax deformation and wing kinematics is poorly understood. We measured the deformation of the dorsal thorax (mesonotum) of tethered flying hawkmoths,Agrius convolvuli, using a high-speed laser profilometer combined with simultaneous recordings of electromyograms and wing kinematics. We observed that locally amplified mesonotum deformation near the wing hinges ensures sufficient wing movement. Furthermore, phase asymmetry in IFM activity leads to phase asymmetry in mesonotum oscillations and wingbeats. Our results revealed the flexibility and controllability of the single structure of the mesonotum by neurogenic action of the IFMs.

Several insects require both olfactory and visual cues during odour-source localisation to successfully locate an odour source. In the male silkmoth, Bombyx mori, detection of the female sex pheromone triggers a programmed walking pattern, starting from a surge (straight-line walking) followed by zigzag walking. Although pheromone-triggered behaviour in silkmoths is well understood, the role of visual cues remains obscure. To address this question, we performed behavioural experiments on tethered-walking moths by recording their locomotion during stimulation with a pheromone and a visual motion pattern (optic flow). The experiments were conducted under open- and closed-loop visual stimuli. We found that the use of optic flow input was determined by the behavioural state of surge and zigzagging. Silkmoths exhibited an optomotor response, which is a behavioural visual response, by turning towards the same direction as optic flow stimuli only during surge, but not during zigzagging. In addition, modulation of the zigzag walking pattern was observed when the moths were presented with biased closed-loop visual stimuli (visual feedback with biased constant optic flow); however, the directional preference mechanism was different from that of the optomotor response. Based on these findings, we suggest that the optomotor response is utilised for course control during straight-line walking, whereas the absence of optomotor response during zigzagging is used to effectively perform the programmed walking pattern. Considering the neural basis of programmed behaviour, we speculate that at least two visual pathways are involved in the state-dependent use of optic flow during odour tracking behaviour in silkmoths.