The Dream Job (or, What Am I Doing Here, Anyway?)
Tromsų, Wednesday June 30, 4:30PM
Image credits: AutoSim Web Site, IAESTE Canada Web Site

''A life without busyness is hardly a life at all!'' - Martin Hansen, AutoSim

I haven't yet told you much about what I'm doing here in Norway. I am here as a trainee of IAESTE, the International Association for the Exchange of Students for Technical Experience.

When I applied to become an IAESTE trainee back last December in Canada, the arrangement seemed too good to be true. For less than $200Cdn - and copies of my transcript and resume - Ginny Arnold and the IAESTE Canada team would do their best to find me a job and accommodations overseas. I would simply pay for a plane ticket and some new luggage.

With the guarantee of a scholarship and stipend next year at MIT, I didn't need to worry about making a windfall this summer. According to Ginny, my salary overseas would easily cover the costs of the plane ticket, the accommodations, and some fun times wherever I was posted.

I was soon presented with a variety of employment opportunities in all four corners of the globe. There were several interesting jobs available in the Scandinavian countries, and that fired off a thought arrow in my brain. Surely an academic conference or business trip will some day take me to southern Europe. A sense of wanderlust should inspire me to visit Australia, the Far East, and the southern continents. But it was entirely possible that this would be my best, maybe only, chance to visit a Scandinavian country. So I filled out the form, requesting a job either in Helsinki, Stockholm, or some place called ''Tromsų.''

The job in Tromsų was vaguely described as something involving programming. It was at AutoSim, a company whose web site showcased their work with virtual reality equipment. I expected I would end up debugging or doing some other such gruntwork, but that seemed like a reasonable price to pay for a job north of the Arctic Circle. I also expected I would be simply ''dropped off'' in Norway to fend for myself, with little or no intervention from IAESTE Norway upon my arrival.

Some of my previous writing is witness to the fact that nothing could have been further from the truth about IAESTE Norway. I received a heartwarming welcome, and now have many good friends here. What's more, the advice and companionship of my new friends have allowed me to explore and discover Tromsų far more rapidly than I ever could have on my own. In exchange for the welcome I have received here, I would have happily worked from 9 to 5 licking the roads clean with my tongue.

Instead, I've been writing artificial intelligence for cars that will drive over virtual roads. Martin Hansen and the enterprising employees of AutoSim have developed and deployed a myriad of virtual reality equipment and software, most of which is being used for driver training purposes. AutoSim develops massive in-house terrain and road network databases that become virtual worlds for their simulators. They also retrofit cars and truck cabins with computer-controlled indicators and sensors that provide input for the simulator.

In a typical simulator, the participant sits in the car, while three projectors aimed towards the front and one in the rear produce a field of view of the virtual scene that spans some 180 degrees. Three-dimensional audio, force-feedback in the wheel and pedals, and hydraulics providing pitch and yaw in lieu of back tires complete the interactive experience. Indeed, the simulator is so immersive that it is not uncommon for new users to get motion sick when they receive mixed signals from the virtual and real worlds.

Typically, over six computers - including a massive Silicon Graphics Onyx box and a sideshow of NT machines - work together to make the magic happen.

Much of the software for the simulator is currently developed by Renault in France. It was originally designed with vehicle testing purposes - as opposed to driver training exercises - in mind. For the past few weeks, I have been taking a close look at the existing software in an effort to determine how I might best contribute to AutoSim over the next two months. The ''traffic'' module that generates the autonomous traffic right now is fairly rudimentary, and is also a challenge to modify.

Because of my background and interests, it seemed to make the most sense for me to begin working on new artificial intelligence for the computer-controlled cars in the virtual world. While examining the current system, I have also completed a survey of existing artificial intelligence techniques and implementations using sources on the Internet and in local libraries.

My destiny was chosen on Monday, shortly after I delivered my first presentation to the team. We concluded that over the next few weeks, I will complete the design of a new artificial intelligence system for the computer-controlled cars in the virtual world. I will then begin implementing the foundation for this system, so that in the months to come, AutoSim can develop their own, more powerful and flexible Autonomous Traffic module.

Driving is a very difficult skill to describe. Driving requires planning, but on a moment-to-moment basis it is very reactive. It demands focussed behaviour, but it also requires facilities for rapid interruption and re-planning. Realistic driving also involves learning - for instance, a driver might learn a faster route between two points on a map. In short, driving requires us to exploit just about every trick in the artificial intelligence book.

As AutoSim's John Markus noted in his thesis, developing artificial intelligence for the driving simulator requires us to work towards two often-conflicting goals. We wish to create vehicles that are realistic enough to enhance the driving simulator, but we also want to set up scenarios that require the student to experience specific situations. Equally important are two technical issues: we want to rapidly and intuitively create new ''scenarios'' for the student to encounter, and we also want to easily modify and debug the artificial intelligence code.

So, how do I enjoy this job? The people reading this who know me best have realized by now that I've just stumbled into my dream job. I get to research and develop the things that most interest me - and get paid for doing it. During the days, I've been reading, researching, absorbing the sorts of things I intended to dive into before reaching MIT, simply because they fascinate me. I wanted to sit on the fjords and quietly absorb. I'm sitting on a rooftop overlooking a fjord, doing just that.

My artificial intelligence proposal for AutoSim has been heavily influenced by Dr Bruce Blumberg's work, which he outlined in his thesis ''Old Tricks, New Dogs: Ethology and Interactive Creatures.'' Dr Blumberg, who will be my supervisor next year at MIT's Media Lab, has developed and implemented a technique for artificial intelligence based on ethology, the study of animal behaviour. I stumbled across his writing shortly after finishing The Legend of the Greasepole, a project that involved a remarkably similar - albeit simpler - artificial intelligence technique that I developed on my own.

Perhaps, in my virtual worlds, I will have a Synthetic Rob lick the road clean with his tongue, because I don't know what I've done to deserve this. To be sure, not everything is roses here: several of the IAESTE trainees have been relegated to what they describe as primarily tedious labwork. We all are enjoying Norway and Tromsų, but I seem to have been particularly fortunate with my job. My employer has granted me the freedom to work independently, and that has only inspired me to want to produce something great in the short time I have left here.

Well, the short time I have left here before I attend MIT, anyway. I have already vowed to return to Tromsų at least once more. After seeing the Arctic Lights exhibit at the planetarium last night, I resolved to visit this city again to see the arctic winter and the Northern Lights for myself. As I originally suspected, this is the sort of town that inspires a deep sense of loyalty.

So now you all know a little bit more about who I am, and what I'm doing here. Over the next two months, while I'm writing these pages and exploring Scandinavia, I'll also be striving to begin the implementation of an artificial intelligence. I promise that, within reason, I'll keep you posted. And if you're ever in a position to accept an IAESTE traineeship or visit Tromsų - you know what my advice would be.


Not only is tomorrow the anniversary of Legend of the Greasepole, but it is also Canada Day. This will be my first Canada Day in four years not spent in our Nation's Capital, Ottawa. Each of my last three Firsts of July was, in some way, a turning point. Of course, I wouldn't dare to expect anything that profound tomorrow, but I certainly intend to celebrate! Expect a story, pictures and some more crazy Tromsų goodness!