James T. Watson

Control of obstacle climbing in the cockroach, Blaberus discoidalis. I. Kinematics

Abstract. An advantage of legged locomotion is the ability to climb over obstacles. We studied deathhead cockroaches as they climbed over plastic blocks in order to characterize the leg movements associated with climbing. Movements were recorded as animals surmounted 5.5-mm or 11-mm obstacles. The smaller obstacles were scaled with little change in running movements. The higher obstacles required altered gaits, leg positions and body posture. The most frequent sequence used was to first tilt the front of the body upward in a rearing stage, and then elevate the center of mass to the level of the top of the block. A horizontal running posture was re-assumed in a leveling-off stage. The action of the middle legs was redirected by rotations of the leg at the thoracal-coxal and the trochanteral-femoral joints. The subsequent extension movements of the coxal-trochanteral and femoral-tibial joints were within the range seen during horizontal running. The structure of proximal leg joints allows for flexibility in leg use by generating subtle, but effective changes in the direction of leg movement. This architecture, along with the resulting re-direction of movements, provides a range of strategies for both animals and walking machines.

Leg kinematics and muscle activity during treadmill running in the cockroach, Blaberus discoidalis: I. Slow running.

Abstract We have combined high-speed video motion analysis of leg movements with electromyogram (EMG) recordings from leg muscles in cockroaches running on a treadmill. The mesothoracic (T2) and metathoracic (T3) legs have different kinematics. While in each leg the coxa-femur (CF) joint moves in unison with the femur-tibia (FT) joint, the relative joint excursions differ between T2 and T3 legs. In T3 legs, the two joints move through approximately the same excursion. In T2 legs, the FT joint moves through a narrower range of angles than the CF joint. In spite of these differences in motion, no differences between the T2 and T3 legs were seen in timing or qualitative patterns of depressor coxa and extensor tibia activity. The average firing frequencies of slow depressor coxa (Ds) and slow extensor tibia (SETi) motor neurons are directly proportional to the average angular velocity of their joints during stance. The average Ds and SETi firing frequency appears to be modulated on a cycle-by-cycle basis to control running speed and orientation. In contrast, while the frequency variations within Ds and SETi bursts were consistent across cycles, the variations within each burst did not parallel variations in the velocity of the relevant joints.

Design and simulation of a cockroach-like hexapod robot

This paper describes the design and simulation of a cockroach-like hexapod robot which is under construction for the purpose of testing control principles which are being extracted from the cockroach. The cockroach was chosen because of its remarkable running and climbing capabilities and because much is known about its biomechanics and control. The robot is designed with five, four, and three degrees of freedom in the front, middle and rear legs, respectively, to permit it to mimic the different functions of cockroach legs. Pneumatic cylinders actuate each joint and provide opposing muscle-like forces to actuate the joints. Pulse-width modulation controls the actuators with the necessary smoothness and precision. A dynamic simulation has been developed to predict loads on the structure and the required joint torques.

Control of climbing behavior in the cockroach, Blaberus discoidalis. II. Motor activities associated with joint movement

Abstract. Deathhead cockroaches employ characteristic postural strategies for surmounting barriers. These include rotation of middle legs to re-direct leg extension and drive the animal upward. However, during climbing the excursions of the joints that play major roles in leg extension are not significantly altered from those seen during running movements. To determine if the motor activity associated with these actions is also unchanged, we examined the electromyogram activity produced by the slow trochanteral extensor and slow tibial extensor motor neurons as deathhead cockroaches climbed over obstacles of two different heights. As they climbed, activity in the slow trochanteral extensor produced a lower extension velocity of the coxal-trochanteral joint than the same frequency of slow trochanteral extensor activity produces during horizontal running. Moreover, the pattern of activity within specific leg cycles was altered. During running, the slow trochanteral extensor generates a high-frequency burst prior to foot set-down. This activity declines through the remainder of the stance phase. During climbing, motor neuron frequency no longer decreased after foot set-down, suggesting that reflex adjustments were made. This conclusion was further supported by the observation that front leg amputees generated even stronger slow trochanteral extensor activity in the middle leg during climbing movements.

Combined intracellular stimulation and high speed video motion analysis of motor control neurons in the cockroach.

A complete understanding of motor control circuitry requires detailed analysis of the behavior produced by the circuitry as well as the connections between individual neurons. A technique is described for combining high speed video motion analysis of leg movements in the cockroach with electrophysiological techniques such as intracellular stimulation/recording from central neurons and EMG recording from leg muscles. Using a restrained preparation, we have quantified leg movements evoked by intracellular stimulation of individual motor neurons and local interneurons. By incorporating motion analysis into the recording paradigm, the transfer functions from electrical activity to movements can be derived. Because distinct and characteristic responses to single and multiple action potentials are seen in slow, intermediate, and fast motor neurons, it is often possible to identify the motor neuron targets of local interneurons. The ability to analyze movement in any plane is especially useful in situations such as blind neuropilar penetrations, where a more restricted motion transducer arrangement may not be in register with the impaled cell. In addition, it is possible to record and analyze such complex phenomena as coordinated movements in multiple joints produced by local interneurons and reflexes produced by proprioceptive feedback due to activity of one motor neuron.