Jake Laser had wanted to build something for his father since he was eight. His first attempt at "bionic legs" was made of cardboard. A few decades later, the budget has increased slightly: racing seat, harness, Bugatti-ish blue trim and an industrial wheeled quadruped robot.
The result looks like the kind of idea you are supposed to sketch at 3 a.m. before deciding never to build it.
He built it.
His father lives with multiple sclerosis and has used a wheelchair for years. In the video, the problems he describes are painfully ordinary: gravel, snow, a small curb, low ground clearance, getting back into the chair when exhausted. You do not need a cliff to make a place inaccessible. A few centimeters can do the job perfectly well.
That is where the project gets interesting. Jake is not trying to make a better conventional wheelchair. He changes the mechanical base entirely.
Going back to legs because wheels get stuck
The robot combines powered wheels with articulated legs. On simple ground it rolls. When the terrain turns bad, the legs can change body height and step over obstacles or stairs.
The machine is a Unitree B2-W, not the smaller Go2-W that several secondary articles have called it. It is an industrial quadruped weighing roughly 85 kg. Unitree advertises up to 120 kg of standing payload and more than 40 kg while moving depending on configuration, and markets it for stairs, slopes and rough terrain. The company also notes that some functions require human intervention or secondary development and warns against dangerous modification.
That warning belongs right next to the spectacular footage.
The robot in the video was not designed as a medical device. Jake says so himself. During testing, the machine still falls over sometimes. So he tests without his father, tunes the software to carry a passenger, adds a harness and tries again on steps, a ladder, rocks and finally a staircase.
Even after the final demonstration, he is clear about the limit: a few spills during development would be unacceptable in a real product.
This is a prototype. A very strange one, but still a prototype.
The revealing hack is not necessarily the robot
Visually, the legs demand all the attention. I find the seat almost more interesting.
Jake chooses a racing seat, wraps it in blue suede, adds panels, headlights, chrome, underglow and even spinning wheel covers. Objectively, none of the spinners help it climb stairs.
But his father is not being handed some beige technical aid that happens to be vaguely medical. He gets a machine they both think is funny and desirable.
It would be silly to derive a universal theory of inclusive design from one YouTube video. Still, there is something useful in the shift: an assistive object does not have to hide its mechanics or ask the person using it to be grateful that it works. It can also have style because the person using it likes style.
Dignity sometimes comes through very serious decisions. Apparently it can also involve chrome headlights.
The prototype mostly shows which building blocks have changed
Jake says he considered exoskeletons before dropping that route for this immediate project. Too complex, too heavy for one person to develop. Instead he starts from a robotic platform that already handles balance, locomotion and obstacle crossing, then focuses his work on the interface with a human.
That pattern keeps becoming more interesting for makers: stop building the foundational machine and build the misuse of an existing one.
The motors, control stack, sensors and much of the locomotion already exist as a product. The project shifts toward mounting, ergonomics, software, testing and adapting the machine to one specific person.
That does not make fabrication easy. It makes something imaginable for a small team that would have required a full laboratory not that long ago.
That is probably the right way to look at this four-legged chair. Not as "the future of wheelchairs," which is far too large a claim for the evidence on screen. As a maker proof: the parts needed to explore that future are beginning to leave research labs.
For now, they still fall over sometimes. That is a fairly clear boundary.