A motorbike-racing robot | Hacker News Reader

A motorbike-racing robot

bbc.com

62 pointsgroodt47 comments

As a qualifier, missing by 30 seconds on a 85 second lap is not very close.

But a 1:55 at Thunderhill would be consider a fairly decent amateur time on a 600cc bike so they are definitely making progress.

As long as they have the robot sitting there static, they're going to be hard pressed to beat a human. There's a lot more going on on a bike than just fiddling with the throttle and steering

“Motobot had to be taught how to balance the bike at lean angles from zero to over 50 degrees, at speeds between 5km/h and over 200 km/h."

Article doesn't say, specifically, but it must be doing something other than just sitting static to achieve this?

[edit] https://www.youtube.com/watch?v=4dUb5T7DU-Q Bike has outriggers!

I’ve found countersteering to be a very effective way of adjusting the angle of a motorbike while riding. You basically push the handlebars in the ‘wrong’ direction and the bike just leans over.

It’s scary when you first try it going down a motorway (it’s less sensitive at higher speed), but very effective indeed, and doesn’t require moving your centre of gravity.

I don’t pretend to be a professional racer though. They do seem to lean a lot.

PS. You can do this on a bicycle too, but the reduced momentum makes it much more sensitive.

All steering after ~15mph is countersteering. There's loads of very cool techniques involved in motorcycle racing, well outlined in Keith Code's book Twist of the Wrist 2, although some appear to be less useful nowadays, such as the speed of reaching lean angle versus the actual angle...

You do this when riding a bicycle too

Counter steering tips the bike into a corner. Some form of lateral force from the front tire and gravity from the lean angle are responsible for the bike turning. In modern racing the limit on the speed you can carry through a corner is the force between the tire and the road surface. This is proportional to the size of the contact patch between the tire and the road.

Modern traction control systems have recently surpassed humans in being able to maintain a maximum amount of power to the rear wheel without making it spin out. Previously more lean usually ment a shorter line through a corner but was balanced by the need for accurate throttle control, leading to the rider being tucked in close to the bike with one knee down. With the new systems the aim has become to stand the bike up again as quickly as possible to maximize the contact patch allowing the traction control to put more power through the back wheel and accelerate the bike faster. This has changed the entire style of riding for professional racers, leading them to hang further out from the bike often touching both elbow and knee to the ground, to try and maintain the same center of gravity of the bike and rider but with the bike in a more upright position creating a larger contact patch and allowing the traction control to accelerate harder.

I wasn't aware motorbike racing allowed the use of traction control.

I know as a fan of F1, that fans generally look down on traction control, is that the same in motorbike racing?

Does MotoGp use traction control systems?

Yes, MotoGP uses traction control. I can't speak for anyone else, but I don't really care. Like F1, it's a prototype class, so results are a combination of the rider and the machine.

Until the last couple of years, the manufacturers all used their own electronics, but now all the bikes must use Magneti Marelli electronics and must figure out how to use those APIs to control spin and throttle. Last season, Yamaha couldn't get this right with its chassis, and its riders couldn't make the tyres last the race with decent grip.

Yes, MotoGP allows traction control. Like all technical innovations it has led to a change in riders priorities and skills. Some people dislike it, but overall it's not accepted I'd say. Personally I've found it interesting to watch riders adapt there riding style to the new technology.

That's how you balance a recumbent bike. It takes some getting used to but after a while it becomes just as much second nature as shifting your COG.

Countersteering is a natural thing isn't it? I don't ride motorbikes very often, but every time I do, countersteering just comes naturally.

A side point, riding a motorbike involves counter steering, just some people dont know they are doing it. For the latter this can be dangerous as when you hit an emergency situation and your brain goes from a relaxed state and thinks 'what do I do' you try to steer away, and end up heading more towards the object you want to avoid.

So if you do ride a motorbike and dont know you are counter steering I'd recommend googling 'conscious counter-steering' and practising this. It could save you on the next bend.

The article is wrong. There is no balance. The bike balances itself. Check youtube for the countless clips of bikes riding on forever after the rider falls off. At racing speeds you really can do all the direction control via steering. You wont be the fastest but you will get across the line.

Motorcycle tires are getting so good that lean angles are limiting the weight-shifting advantage. Look at the 1980s and you will see guys hanging low off thier bikes dragging knees. Todays racers can lean the bike so far that they drag elbows. There literally isnt enough room to hang. So the robot isnt at too much of a disadvantage.

It does look like it just sits there static. I actually saw this robot/bike at CES earlier this month (just the showcase, I didn't see it in action). The robot looked like a 'dummy'. As if there's no real practical purpose of the robot being there! They might have as well just fitted all the sensors on the bike itself, but that probably won't be good for PR.

The humanoid form is just for show. You need a handful of actuators to work the control and a way to power them. Throttle is already servo-actuated on the stock bike. The robot will have the advantage in mass and mass distribution.

What I do think the bots will have a hard time doing is replicating the intuition that people like Rossi and Marquez have. Understanding what parts of the track have the most grip, braking/accelerating/cornering on the ragged edge of traction. And that's only half of what's happening in race conditions -- you have to contend with a grid full of other riders taking offensive lines and passing only inches away.

A robot motorcycle might be able to take on GP-level riders at Thunder Hill in a time trial in a few years, but it's hard to imagine the same thing happening at, say, the Isle of Man TT, or anything approaching race conditions.

> you have to contend with a grid full of other riders taking offensive lines and passing only inches away

It'll be interesting to see what happens if we have robots and humans racing together. On one hand they will be programmed to win, but on the other hand, harming humans is a big no no.

It's likely that this will happen before we have true AI, so it'll be the human overlords (i.e. team mangers) who make that moral decision. Given some of the things racing teams have done before to win, it could be quite concerning...

Agreed. Human racers hang off the bike while cornering, so the bike itself doesn't need to lean so far. This keeps the suspension geometry within a more reasonable range, and keeps a larger contact patch between the tyres and the track. I doubt it's possible to get close to the best human riders unless you can move the centre of mass around laterally.

Footage:

https://global.yamaha-motor.com/showroom/motobot/ep3/index.h...

It is using GPS for navigation at the moment. I wonder if the reduction in Lidar size and advancements in cameras would mean that in theory these things could be raced in groups. That would be cool.

Nice. Now we just need the robot to ride with 1 hand and reload a shotgun with the other.

Why not just have three hands? And why have a shotgun when a laser is better!

I don't think James Cameron thought that far

It's an infiltration model designed to pass as human until it reaches its target.

Then fight this one: https://www.youtube.com/watch?v=6LjKfs7MjYE

There's a lot more to racing than simply being able to go fast around a track solo, but this was a fabulous piece of work from Yamaha. This was last summer, I think, so it's a bit late for the Beeb to be reporting on it.

Soon, I imagine that the bike will also start to be tailored to the robot in the way that it's tailored to a human (two humans, actually, Rossi and Viñales).

Hello guys,

I am finding it hard to understand how they managed to model the system(i.e bike). I am somewhat familiar with control theory(did some undergrad courses on them). But this system seems like a non-linear system and coming up with a mathematical model seems like a herculian task to me.(I once modeled a conical tank and it was not quite easy)

Could someday provide me any idea about what they have done here ? How they modeled this system ? What control algorithms they used here ?

Any pointers might be helpful as well.

P.S - Haven't read about developments in control theory in quite some time now. So forgive me if my comment is quite basic.

Thank you.

"We decided that physically attaching Motobot to the bike wasn’t cheating because its hands had to grip the handlebars"

Come on guys. This is just like Watson's finger or AlphaGo's precomputation. It is cheating obviously!!

If it's cheating it's not very useful in gaining an advantage. Hanging the upper body out and forward is a core component of the contemporary winning motogp riding technique.

Of course it's cheating. The humans have to deal first and foremost with not falling off. That is very costly. They do a lot of actions and calculations to prevent this. The robot has unfair advantages like not caring about getting hurt, and being assured it can't fall off. Given a human 100% sure of that I bet they can also go 200 km/h or even 400!

> AlphaGo's precomputation

What are you talking about, training the Go neural net?

the robot should get back on the bike when it falls off it, and then you can tell it to experiment with leaning etc.. and just leave it for months on end.

A video of this robot: https://www.youtube.com/watch?v=BjZPvXKewFk

How soon will Doordash have this technology?

> “Why a motorbike?” ponders Hiroshi Saijou. “Because it is very difficult to do, and it had never been done before.”

I feel compelled to point out that it has in fact been done before. There was an autonomous motorcycle -- Ghostrider -- which was entered into the '04 and '05 DARPA Grand Challenge, and now rests in the Smithsonian. [1]

[1] http://americanhistory.si.edu/collections/search/object/nmah...

That's a modified motorcycle. The one in the article is a normal motorcycle with a robot sitting on it.

Putting the actuators on the motorcycle is the easy part. From a control perspective, it doesn't much matter whether the actuators are packaged into a humanoid form or not.

You have much more precise control over things like throttle position if you're controlling a fly-by-wire throttle electronically compared to actuating a physical throttle mechanically.

In principle I agree with you, but I think the inaccuracies will stack up quite quickly.

Getting a motorcycle to change direction is more than just actuating a control, body position, centre of gravity and using the legs to push down on the pegs are all things racers do.

This is true! Nice username, btw!

I think some of this comes down to how closely you have to mimic a human rider. Does the robot have to weigh as much as a plausible GP rider? If not, then the robot plainly has a mass advantage. Would that be counteracted by the robot's inability to shift mass and drag knee/elbow like a Rossi or Marquez can? I actually don't know.

I knew my username would be exposed at some point :)

The robot will absolutely have to weigh as much as a Marquez - just look at the disadvantage that a super-light rider like Pedrosa has on some tracks - he just can't get enough heat into the tyres to generate enough grip.

Humans adapt quickly to changes in the bikes and variations in tyres and surfaces, the robot however, could in theory be directly linked up to the electronic systems to get direct rather than physical feedback (I'd call that cheating)...

Do any of the two contain (hidden) gyroscopes? Because I feel that would be cheating.

Why would that be cheating? Your body uses a pretty clever system in the inner ear with little chalk particles that touch hairs to tell you which way your body is oriented, it can also detect acceleration in different directions. The effect is much like a gyroscope, except that it does not use a wheel.

There are two kinds of gyros. Small ones for sensing and large ones for actuation. I think it's obvious the poster meant the latter. Anyway, for sensing, you nowadays use laser gyros and mems accelerometers instead of mechanical gyros for sensing.

> I think it's obvious the poster meant the latter.

It wasn't obvious to me, but now that you mention it, yes, that would be a neat little trick to pull.

They both do, and neither make any pretense around hiding them. In fact, the stock R1M has gyros without a rider of any sort, as part of its traction control system.

Reply on news.ycombinator.com