- Biomechanical energy harvesting can provide an attractive alternative for powering portable electronics, such as laptops and mobile communication devices. This unique type of energy is particularly advantageous to those who do not have direct access to the utility grid for extended periods of time. This paper presents an innovative biomechanical energy harvesting system based on the regenerative braking concept applied to the human natural motion. To determine the optimal braking profile, previous studies used an offline procedure based on constant external load to determine the optimal braking profile. The new concept of this paper continuously optimizes the maximum amount of energy that can be extracted during human motion while minimizing the subject's effort (metabolic rate). This is achieved by an energy harvesting system equipped with a programmable braking profile and a unique power extraction algorithm, which adaptively changes the braking profile to obtain the optimal ratio of energy to effort. These are facilitated by a brushless dc generator that is connected to boost converter. A digital current-programmed control of the boost converter enables an adaptive torque variation according to biofeedback (the measure of effort) and electrical feedback. This paper focuses on the human knee joint as the energy source, since most of this joint's work during level walking is negative (muscles are acting as brakes). Since this paper is preliminary and more oriented to the novel concept of adaptive profile and optimal power extraction, the operation of the energy harvester is demonstrated on a full-scale laboratory prototype based on a walking emulator. Initial experiments on human subjects have been initiated and are also reported. The results exhibit ultimate power extraction capabilities as well as adaptation to the walking pattern.