Parent comment was about a simple trigger "similar to pacemakers", it was a reply to that. It was never implied that it's not possible with the current technology

That press release links to a journal article in Nature. I can't read the Nature article because it's paywalled, but the same group published an article in Brain in 2014 where they said "We propose that the functional state of... motor neurons was modulated by the epidural stimulation, presumably driving them closer to their appropriate activation threshold, enabling intentional movement." [1] Basically, they're saying that the spinal cord was a bit blunted after the injury. Movement signals were fizzling out as they travelled through the injured area. The stimulator is giving the injured region of the spinal cord some extra juice so that it propagates motor signals further down the body. However, it isn't actually creating a signal de novo - that's still coming from the brain. As for the latest Nature article, based on what I can see from the abstract, it seems like it is an update where they used a computational method to decide where to place the stimulators and with which strength to stimulate. But it's still the same basic idea.

Another thing to note here. In the 2014 article, they mention that they only tried this treatment in patients who had an intact sensory pathway and an injured motor pathway. Kind of speaks to what I was mentioning earlier - the brain can't coordinate movement unless it gets feedback from the muscle. I would presume that this stimulator therapy can't help with the sensory component, which is why the researchers limited the study to patients with sensation.

[1]: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3999714/

Somehow I did manage to access the article, and while I didn't quite understand exactly what they did, it does go beyond dumb amplification.

> Neurostimulation platform. Biomimetic EES requires the delivery of concurrent stimulation waveforms that are turned on and off with a precise timing4,8,12,21. Moreover, many activities necessitate adjustment of stimulation parameters in closed-loop via wireless links. To support these features, we upgraded the Activa RC implantable pulse generator (IPG) with wireless communication modules (Supplementary Fig. 1). This neurostimulation platform supported real-time updates of EES frequency, amplitude and timing from up to 10 stimulation waveforms8. The new paddle lead was interfaced with the Activa RC, which was implanted in the abdomen. We also developed a new software operating through touchscreen interfaces to enable the rapid configuration of activity-dependent stimulation programs (Fig. 4c). To simplify these configurations, wireless recordings of kinematics and muscle activity are displayed in real time, concomitantly to EES waveforms (Supplementary Fig. 1 and Supplementary Video 2).

Regarding Cerbral Palsy, it is... Cerebral, meaning that the peripheral part of the sensory and motor pathways should be intact. So what you would need to do is develop a controller interfacing with them both. That may not be trivial, but I do think it could be done by aplication of known tech and principles.