The Tokyo Motor Show always has a collection of futuristic concepts and prototypes alongside production motorcycles. And 2017 is no exception. The Yamaha MOTOROiD AI motorcycle is part of a range of electric and semi-autonomous ideas on display from the Japanese manufacturer.
Images of the Yamaha MOTOROiD AI motorcycle have been released in the past, and now there’s actually an ‘experimental proof-of-concept’ built. Sadly, there’s not a lot more detail about what it actually can do.
Yamaha claim it ‘explores creating new forms of personal mobility in which the rider resonates harmoniously with the machine’, and that ‘ it is capable of recognising its owner and interacting in other capacities like a living creature’.
We do know it has a wet weight of 213kg, and a lithium-ion power supply for the rear-wheel hub motor. And that it’s 2,060mm long, 600mm wide, and 1,090mm high.
Yamaha MOTOROiD AI Motorcycle Capabilities?
Artificial Intelligence and self-driving vehicles are dominating news about the future. The main thrust so far has been autonomous cars developed by the likes of Google, Tesla and Uber. But self-driving lorries and ships are also being developed. And it makes sense that similar technology will come to motorcycles eventually. Whether or not it’s a good thing depends on the final product, but you don’t need to panic quite yet.
There’s a lot of debate whether artificial intelligence can replicate and surpass the human brain one day. It’s referred to as the technological singularity, and scientists, programmers and science fiction writers have been talking about it for decades.
What is happening is that there are some approached to artificial intelligence which are making big leaps forward. And without going into ultra-techy mode, there’s basically two approaches. One is to essentially teach an artificial intelligence rules to automatically follow, and that seems the most likely route for the Yamaha MOTOROiD AI motorcycle at the moment.
So for example, it may detect the owner through facial recognition, a thumbprint or a key fob and unlock itself. Or maybe it could fold up the rear swingarm when parked, and lower it when the owner appears to save space. Or turn on the lights, etc.
And when riding, ‘resonating harmoniously’ may involve a heads-up display (HUD) for a linked crash helmet. That’s something that was displayed in the BMW R1200RS ConnectedRide we featured from the Connected Motorcycle Consortium Conference. It’s worth remembering that the association of manufacturers was founded by BMW, Honda and Yamaha, and that a Yamaha Tracer 900 with a similar system was also on display.
The other approach to artificial intelligence is using neural networks and deep learning to become a self-taught expert in whatever it is created to do. For example, Google’s DeepMind company originally built an AI to become an expert at playing the Japanese board game Go by showing it more than 100,000 games to get it started.
The newest version of the AlphaGo AI was simply given some basic rules of the game, and then left to play itself and learn. Both were able to beat the best Go players in the world, but the new version spent 3 days playing millions of games with itself, and then beat an 18-time human champion 100 games to 0.
So you’d think that a motorcycle with a self-learning AI would be here soon, and the days of having any control over your motorcycle are gone…
But you’ve got a fair bit of time yet. The starting point for that board-game playing AI was the basic set of rules. And that approach was also used by OpenMind, a company backed by Tesla boss Elon Musk, to beat professional human players in the much more complex Dota 2 videogame. And the self-learning AlphaGo bot beat the rules-taught version when they were pitted against each other.
The problem comes when you add more and more complexity. When Facebook put a self-learning bot against rules-based home-built versions by hobbyists in a Starcraft 2 competition, Facebook lost.
And the reason is that the number of positions on a Go board is a 1 followed by 170 zeros. The estimates for Dota 2 and StarCraft is that you’d need at least 100 more zeros.
It’s why AI companies are now breaking down those games into smaller mini-games and tasks to make it more manageable.
And it’s also why a completely autonomous motorcycle riding AI is likely to take a long while before it’s ready to get close to release. Human life is much more complex and unpredictable than a board game, and even the complexities of Starcraft or Dota 2 pale next to the range of things that could happen on your way to work.
It’s the difference between your car automatically parking, and your car automatically taking you on a trip to Scotland for the day. And motorcycles add more challenges due to the lack of space for cameras and radar, the lack of storage for the computers required for processing and batteries needed to power it (seen in the canisters on the side of the Yamaha MOTOROiD AI motorcycle), and also a greater amount of calculations needed for cornering, particularly at higher speeds.
So yes, the Yamaha MOTOROiD AI motorcycle looks cool. And it’s likely to show where we’ll get to eventually with powered two-wheel vehicles. We’re unlikely to get fossil fuelled motorcycles when sales of new petrol and diesel cars are banned. And AI will increasing play a larger and larger part of our riding experience. But it’s going to be a long time before it’s in a position to take over. And as long as you can turn it off for track days or a Sunday blast, it’s not necessarily going to be a bad thing when it does…