The development of wearable devices and remote sensor networks progressively relies on their increased power autonomy, which can be further expanded by replacing conventional power sources, characterized by limited lifetimes, with energy harvesting systems. Due to its pervasiveness, kinetic energy is considered as one of the most promising energy forms, especially when combined with the simple and scalable piezoelectric approach. The integration of piezoelectric energy harvesters, generally in the form of bimorph cantilevers, with wearable and remote sensors, highlighted a drawback of such a configuration, i.e., their narrow operating bandwidth. In order to overcome this disadvantage while maximizing power outputs, optimized cantilever geometries, developed using the design of experiments approach, are analysed and combined in this work with frequency up-conversion excitation that allows converting random kinetic ambient motion into a periodical excitation of the harvester. The developed optimised designs, all with the same harvesters’ footprint area of 23 × 15 mm, are thoroughly analysed via coupled harmonic and transient numerical analyses, along with the mostly neglected strength analyses. The models are validated experimentally via innovative experimental setups. The thus-proposed ϕ = 50 mm watch-like prototype allows, by using a rotating flywheel, the collection of low-frequency (ca. 1 to 3 Hz) human kinetic energy, and the periodic excitation of the optimized harvesters that, oscillating at their eigenfrequencies (~325 to ~930 Hz), display specific power outputs improved by up to 5.5 times, when compared to a conventional rectangular form, with maximal power outputs of up to >130 mW and average power outputs of up to >3 mW. These power levels should amply satisfy the requirements of factual wearable medical systems, while providing also an adaptability to accommodate several diverse sensors. All of this creates the preconditions for the development of novel autonomous wearable devices aimed not only at sensor networks for remote patient monitoring and telemedicine, but, potentially, also for IoT and structural health monitoring.