Vibration energy harvesting is a promising step towards a more sustainable society. The world is getting more and more connected through electronic devices, which all require electrical energy. A battery can deliver electrical energy; however, such a battery needs to be replaced or recharged; this is where energy harvesters become interesting. Energy harvesters convert energy from ambient sources to electricity; this source can be, for example, solar or thermal energy but also vibrational energy. Extensive research has been done in vibrational energy harvesting so far. The subject of human motion energy harvesting is increasing interest. An energy harvester for human motion can be used to power health monitoring devices. However, there is one significant problem; human motions are dominantly low-frequency with high amplitude motions, while energy harvesters tend to work better on high frequencies. A limiting factor for successful experimental research is the equipment. The lack of sufficient stroke, controlled and low-frequency excitation impede research regarding human motion. In this research, a new test setup is developed for experimental research. A linear air-bearing stage is used to reproduce the human motions with an amplitude up to 500 mm. The air-bearing stage has an incremental encoder to ensure a precision of at least 20 µm. This stage may be used for many different testing situations ranging from vibration testing to impact testing. The stage can reproduce motions that were impossible to reproduce with a shaker, for example. The research is expanded with a nonlinear oscillator to assess its performance on large amplitude motions. A transducer can be attached to the oscillator to transduce the vibrational energy into electrical energy. By using an oscillator, the frequency is increased, causing a higher energy output at low frequencies. The dynamics of the nonlinear oscillator are numerically calculated. For which a new method is proposed to simulate the bouncing behavior in the spring numerically, called the bounce loss coefficient. The new method shows better results than the traditional model used to simulate the bouncing behavior (coefficient of restitution). The numerical model is experimentally verified on the newly developed testing setup. It was shown that using a new model for the bouncing behavior, the dynamical behavior of the nonlinear oscillator can be reproduced when excited at a large amplitude motion.