There is theoretically a way to use the roughness of the terrain (so the bike going up and down as it rides) to store that energy and then use that for propulsion. When finding a rough spot, the bike trades forward momentum for a vertical impulse, so it might be possible to transform a part of this back to horizontal. I don't have a basis to calculate this, but my intuition tells me accounting for loses this energy must be very small and very difficult to capture, to the point of almost worthless. Think of a wave energy generator, but on a calm day with waves almost flat.
The article doesn't claim that though ("10kg active weight [...] generate a moment ranging from 7.3Nm to 7.8Nm"), so this is either bad reporting or a perpetual motion scam. My money is on the latter specially seeing the "infinite power assistance" label on the video, would be happy to be proven wrong though!
Fun fact: In 1775 the Royal Academy of Sciences in Paris stopped accepting proposals concerning perpetual motion.