A musculoskeletal model of pedaling was developed which consisted of the forward dynamical equations of motion for two legs, with muscles, coupled via cranks to a flywheel with frictional load. This model of the entire system allowed decomposition of the equations of motion to find the contributions of each muscle force to the mechanical power in each segment, including the crank inertia. Controls (muscle excitations) to produce simulations of pedaling were determined using an algorithm which could be structured to optimize different goals, such as pedaling as fast as possible, or minimizing metabolic energy consumption over a cycle. These simulations show how muscles must be coordinated in pedaling, and provide a baseline for comparison with experiments.

A simulation of maximum-speed pedaling, where the goal of the task is clearly-defined, was found to agree with experimentally measured pedal reaction force, kinematic, and electromyographic data. Analysis of the energy flow between muscles and segments indicated that some muscles must work in synergy to propel the crank while avoiding kinematically unfavorable configurations, while others can function independently in certain parts of a cycle. For example, while uniarticular hip and knee extensor muscles produce most of the energy required to overcome the load and accelerate the crank, much of this energy is not delivered directly to the crank via intersegmental reaction forces. Rather, the ankle plantarflexor muscles transfer this energy to the crank, and in so doing prevent hyperextension of the knee at the end of downstroke. The biarticular hamstrings and rectus femoris muscles are the primary energy-contributors during the transitions from downstroke-to-upstroke, and vice versa, respectively. Therefore, excitation of these muscles is necessary so that, with the other muscles, energy can be delivered to the crank evenly around the cycle to produce smooth pedaling. Thus, the simplest effective control for pedaling consists of a primitive locomotor strategy which excites two pairs of alternating muscle-groups, where the two groups of each pair alternate with each other in the crank cycle. In this scheme, the uniarticular hip/knee extensor muscles are excited during downstroke, in alternation with the uniarticular hip/knee flexor muscles during upstroke. The other pair consists of the rectus femoris and ankle dorsiflexor muscles, excited through the transition from upstroke to downstroke, and the hamstrings and ankle plantarflexor muscles, excited during the downstroke-to-upstroke transition.

The primitive locomotor strategy can accomodate many task goals and conditions in normal pedaling. Power output to the crank (which determines cadence for a given load) can be selected by setting the overall excitatory drive onto the muscle groups. Subtle shifts in task goal (e.g., energy-efficiency vs. pedaling smoothly) require setting of excitation levels of the muscle groups relative to one another as well as their phasing. Separation of the ankle muscles and biarticular thigh muscles into two independent pairs was found to increase flexibility and robustness of the control, consistent with the need to fine-tune control of the ankle muscles so the feet are properly positioned to effectively transfer energy from the limb segments to the crank. The primitive strategy only has to be minimally reorganized to effect different pedaling tasks. Smooth backward pedaling requires the biarticular thigh muscles to be reversed in phase, consistent with the phase changes seen in backward walking. Non-seated pedaling, where body weight is supported, can be accomplished by exciting extensors, not only in the downstroke, but in the upstroke in lieu of the flexors. In contrast, a strategy simpler than the "primitive" (fewer muscle groups) resulted in significantly impaired movement in all cases. We hypothesize that load- and motion-dependent feedback are utilized to reset parameters in the primitive strategy initially, and only occasionally thereafter.