London CGI firm scores DON'T-HIT-the-PEDESTRIAN test contract The Driving Standards Agency (DSA) has awarded animation firm Jelly a contract for CGI technology. DSA said it met with the SME last October to identify new technologies that could be used to refresh its hazard perception test. The agency started a formal procurement process in January and two early examples of CGI clips were …
"The use of computer-generated images means that we can include situations with vulnerable road users such as children, cyclists, and motorcyclists without risk."
So do they mean that they were putting real road users such as children at risk by placing them in front of the stationary test cars in the simulators? Stupid sentence.
And why does it need a more realistic CGI. It's all about noticing hazards to the driver, whether they are children or motorcyclists should not matter. They could just as well have place coloured boxes on the animation and got the learner to drive around them.
I failed the first time when I clicked as soon as I saw a potential hazard. The second time round the only change to my strategy was to click about three times for each hazard, once when I originally saw it, once slightly later but just before I might have to react and once when I had to react. I might skip some if I was worried about the system picking up on a triggering a clicking too much warning.
In real life the scenario is more complex you might identify a number of hazards that you need to track at different priorities and then situations with each hazard may change and you may have to do something. Having a click represent this doesn't work especially well in my opinion and it felt like I was trying to work out the magic sequence to pass the test rather than illustrate my awareness to hazards.
I'd be interested to know what this firm comes up with.
I had a similar problem when I took my motorbike test a few years ago and bought some practice CDs.
From many years of cycling, I'm very aware of watching for hazards (and, yes, I *know* some cyclists are numpties, but let's not get into that now, ok?) so I would click as soon as I saw something which could turn out to be hazardous.
But the Test wants you to spot a "Developing Hazard" ie you only click *after* something starts actually becoming a danger to you, so I had to slow down my responses in order to succeed!
There's also the problem that the Hazard Perception Test just shows a fixed view ahead, so, for example, there's no situation where the driver approach a T-Junction (from the minor road) and pulls out in front of a biker who they haven't seen because they didn't look properly...
"The second time round the only change to my strategy was to click about three times for each hazard, once when I originally saw it, once slightly later but just before I might have to react and once when I had to react. I might skip some if I was worried about the system picking up on a triggering a clicking too much warning."
That's exactly what they instructions at the start of the test tell you to do. You're meant to click at each stage the hazard changes. One of those changes is the one they're monitoring for.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
