Guy Martin v The Robot Car

[Inertia]

Another question, what about snow. Will the car simply refuse to leave the driveway as it is too dangerous?

Awesome.

"Sorry I can't come into work today, the car ran a risk assessment and, its a no go."

 

[Deleted member 26715]

In very slow traffic where cars push and jostle it would be very easy to just keep pushing past a robot car.

I there is mixed traffic this could happen, but if it was segregated then it should flow far better, they could only ever pull up where allowed not randomly, they could be programmed to merge alternatively, but the issue will be humans that will cause the problems.

 

[Deleted member 26715]

It has to be more complex than just putting braking points in. The conditions, road position could change each time.

Where they actually trying to achieve AI the more I think about it they wasn't it was just an exercise to see how fast a car can go round the track

 

[classic33]

And of course where humans and robot mix there will be the people that know they can get the robot to give way through reckless action, the same way that people do with other drivers now when they perceive them as inferior.

Some manufacturers are only now looking at the moral side of programming the vehicles. Introducing a "moral lever" for the passenger(s) to use to override the programming.

 

[Tim Hall]

Just watched it on the catch up thingy.

What were they thinking? The tea making gadget put the milk in before the boiling water. <tuts loudly>

 

[mjr]

"Top tip" - get with the times. They are now 'life hacks'

Did you print your avatar by filling it with cream?

 

[classic33]

[QUOTE 5061135, member: 9609"]just watched it - didn't seem anywhere as near as advanced as I was suspecting, could only do 100% or 0% throttle ? It was needing many laps to learn an empty track, so could it have coped racing another car and been able to choose the best line for the space available.. How would it have coped if t had started raining, Ai just seemed clueless

a more interesting test would have been to compare Guy against Ai on an unknown track without any practise goes.[/QUOTE]
No AI, just plain robotics relying on a plan of the track pre-programmed into it.

 

[mjr]

No AI, just plain robotics relying on a plan of the track pre-programmed into it.

The two aren't contradictory. If the car controller was modifying its output based on monitoring its environment (calculating throttle setting based on current track position, for example), then that's a type of AI. It doesn't have to use a neural network or similar.

 

[hobo]

Awesome.

"Sorry I can't come into work today, the car ran a risk assessment and, its a no go."[/QUOTE

 

[hobo]

Well you won't have to go to work as you're be sending your personal robot in to do the work for you

 

[Deleted member 26715]

The two aren't contradictory. If the car controller was modifying its output based on monitoring its environment (calculating throttle setting based on current track position, for example), then that's a type of AI. It doesn't have to use a neural network or similar.

Correct me if I'm wrong, but as I see it @classic33 is refering to a system which would send a car around a track so that it accelerates at the same spot, brakes at the same spot, turns the steering exactly the same amount at the same time etc. to make it change that the programmer actually makes changes, shortens braking point, accelerates harder, changes road position etc. But an AI would do this itself & then learn from it's results & modify it's behaviour due to it, for instance it brakes 2M later & ends up on grass so doesn't do that again from that position.

 

[classic33]

Correct me if I'm wrong, but as I see it @classic33 is refering to a system which would send a car around a track so that it accelerates at the same spot, brakes at the same spot, turns the steering exactly the same amount at the same time etc. to make it change that the programmer actually makes changes, shortens braking point, accelerates harder, changes road position etc. But an AI would do this itself & then learn from it's results & modify it's behaviour due to it, for instance it brakes 2M later & ends up on grass so doesn't do that again from that position.

It was programmed to stay within the track limits, with an increase in speed with every run.

It reached and passed the point at which going faster, whilst remaining within track limits was no longer possible, when speed was increased alone.

 

[Deleted member 26715]

It was programmed to stay within the track limits, with an increase in speed with every run.

It reached and passed the point at which going faster, whilst remaining within track limits was no longer possible, when speed was increased alone.

I found it a simplistic system which I'm sure it isn't, but after the slide at the first corner it didn't seem to realise it was not in the right place to execute the next operation hence the gravel

 

[classic33]

I found it a simplistic system which I'm sure it isn't, but after the slide at the first corner it didn't seem to realise it was not in the right place to execute the next operation hence the gravel

Go back to the Scalextric set at the start of the show. Two cars following a slot round a track. Simple to understand.

Get a digital set and you can programme a car to come out "when it likes, travel at a speed and in any lane it likes", the rest then either race that car or stay behind it(Safety Car, with it "deciding" the speed). Things start getting complicated.

 

[mjr]

I found it a simplistic system which I'm sure it isn't, but after the slide at the first corner it didn't seem to realise it was not in the right place to execute the next operation hence the gravel

I thought the problem was that it knew where it was, but failed to react to the lack of grip in the first hard acceleration, so it accelerated too hard again and binned it.

When they were modifying the program, they were changing variables called stuff like scaling_factor which didn't look like simply changing how much it accelerated where, but more like how quickly it attempted to change speed.