DARPA Urban Challenge

[pgrayson]

The largest race in robotics history, with up to US$11,000,000 in cash prizes, is just 49 weeks away.

Participation is open to all ( You have missed the registration deadline but can participate by joining my team ).

This is a rally style race through 60 miles of heavy urban traffic.  Vehicle with the shortest time, without being disqualified wins the prize.  These are full size vehicles in traffic with other full size vehicles.   

The rules are about an inch thick and basically specify that the robot vehicle has to be able to pass a California State Drivers License Exam to qualify to be allowed on the course.

This is my fifth year working with this race.  My team is made up entirely of volunteers and is supported by donations. 

There is a great deal that you can learn about robotics by being on my team as we push the frontier of Robotics knowledge forward. 

If you want to be on my team, I will try to find a place for you whatever your skill level is.  It takes many hands and many minds to move a project like this ahead.  Some tasks are simple, others more complex, there is something for everyone who wants to help. 

Volunteers are needed from now through race day Nov. 3, 2007, a commitment of two hours a week or more is required to qualify for shares of the prize money.  While volunteers may quit at any time, they will not be allowed to work for a competing team.

 Does it sound like fun?  Are you up to the challenge?  Contact me as soon as possible.

[JesseWelling]

I'm kinda curious what kind of programing you guys are doing right now?

Most of my experience comes from embedded systems, but I would also
like the opportunity to learn more about what it takes to make a
fully autonomous AI(I've taken our schools intro to AI). Right now
I'm undertaking my own project to make moblie robot based on
a R/C car. I have experience with Real Time Operating Systems
and general Linux programing.

[Militoy]

The largest race in robotics history, with up to US$11,000,000 in cash prizes, is just 49 weeks away.

Paul, has DARPA officially re-instated the prize money? I hadn't noticed any announcement on the Urban Challenge site yet.

[pgrayson]

JesseWelling,
I can't say much about the programming in a public forum but your background and your current project are directly related to what we are doing in full scale.  Do you want to play on our team?

Militoy,
DARPA's boss said there will be a cash prize of up to several million dollars.  Originally it was 11 million, then it went away, now it is back but the exact amount has yet to be determined.

Cognaut,
Distance is not a problem, we have had 60 people on three continents, spread over 15 time zones working long distance on this project.  Also a pit crew team here in Traverse City, MI who do the final assembly and travel with the vehicle.  Other team members are welcome to attend exhibitions and races we participate in - if they can afford to.

As you can see on my home page, there is a link to the free hour long documentary if you have a high speed connection.  Also if you go to YouTube.com and search for DARPA you will bring up some of the other teams videos.  Some of the other teams home pages linked from the DARPA button on my home page have videos of their vehicles in action.  Nothing is quite as good as seeing what the competition is doing.

[pgrayson]

cognaut,

To answer your first question, I can't tell if you are a robot or not.  But if you are I hope that you are operating at a fairly hight level and are loaded with information.  My conversations with ALICE interfaces to large data banks have been very interesting.  The team could use one of those to let the public and reporters ask the vehicle itself, AV WENDY DARLING about the race (DARPA data), about the team (AI MAGIC data) and about the vehicle itself (AV WENDY DARLING data).

> How do you direct/task your programmers?  That's my skill set.  I'd like to get an idea of how I'd get up to speed and contribute.  I'm sure that you've already got a lot of development "under the hood."

11 new programmers have joined the team and I have lumped them together as a Software Sub Group but they are not yet producing.  Although they say they have read the rules for the race, which they need to design to, they are talking about inventing all kinds of things that are not required by the race and are fretting over problems that are also not part of the race.

To end this I have asked each of them to pick one of the required behaviors from the inch thick rules / specifications and give me a rough block diagram / flow chart of what inputs the behavior requires, what outputs the behavior requires, and roughly what steps there should be in the f(x) program for that behavior.

So far they are still in shock.  Broken down like this the task is very doable but requires actual work.  We shall see who among them is capable of putting on the harness and pulling toward the goal. 

Perhaps with your background you could suggest how to do it better or would like to be part of the team and make it happen? 

If anyone out there would like to audition for the team, you can do so by picking a piece of the driverless vehicle problem and showing me how you would solve it.  An outline, schematic, sketch, or blueprint would be fine.

[Admin]

ive considered in the past using internet collaboration in building robots with people around the world . . . but decided it just wouldnt work due to communication difficulties. you can always have a ring of advisors through the internet (such as this forum), but found the best way to hash out design details is for everyone to be in the same room.

your project is even more complex, with software engineers, electrical engineers, and mechanical engineers all coming from different schools of thought . . . how do you get them to work together through internet alone?

pgrayson, I am curious about your thoughts on this problem, and solutions you have found to it . . .

[pgrayson]

The books I have read on the subject of distributed work force and cottage industry all made it sound like it would be the best of the two worlds.   Later publications and practical experience by large companies who tried to make it work for years have found that it didn't work as well as advertised.  Most companies have abandoned the idea and no longer offer it as an option to their employees.

It turns out to be more difficult to administer than originally estimated.
The people are about half as productive as they would be if they were in the same room.
People who opt for working by internet often find that they miss the social aspects of showing up in person at a work place, even they didn't realize this would be important to them until they tried it for a while.

There are workarounds for these problems.
These work arounds are in addition to the projects work load.

If you are considering organizing a project this way you must ask yourself if your reason for doing it this way is good enough to offset the drawbacks.  Edison found that the most effective/efficient way to get high tech projects done was to put everyone in the same room literally.

In my case the people I need are where ever they are and the hours they can work are what ever is available outside of their regular jobs.  That leaves me with trying to figure out how to make it work.  I am breaking new ground with some of the things we are doing - organization wise. 

A Virtual reality collaborative workspace is something that might be worth looking into.  My work in VR was a long time ago and things have improved a lot since then. 

Any suggestions about how to make it work better?

[JonHylands]

I'd better focus on my own work because I think JonHylands is catching up to me.  hehe.

Catching up?

I'm going to have a walking biped robot using the motion control part of my AI brain by late spring... 

- Jon (busy writing a Squeak/gumstix driver for the Bioloid serial bus...)

[Militoy]

Paul, from experience, I have to concur with you that trying to run a “distributed” team, and keep it on task, is an enormous challenge to overcome. We tried at our company for awhile to run our Manufacturing processes out of one plant, and our Engineering out of another (just a few miles away), with less than satisfactory results – the logistics just slowed everything down. The Urban Challenge team I work on is basically concentrated in one city, but with several important elements located in another state. Our main solutions are extensive travel expenses, and dual test vehicles. As far as pulling your team together and organizing tasks – I believe that in order to be competitive in this next challenge, you will have to be actively testing and integrating your software and sensors 24/7 by the end of this year.

[pgrayson]

cognaut,
I know what you mean.  Luckily you can use some of your screen time to look for a job, postings and apply buttons are now common.  Of course just walking into a business near by and saying to the person in charge that you want to work there works a lot better than resumes and interviews.

JonHylands,
Will your biped be something you will market and sell?

Miltoy,
Team members travel is at their own expense here.  One guy wanted to work for us if we flew him from Europe twice a month - we had to pass on that.  We have multiple vehicles and oddly, clusters of team members in some towns.  Each joined individually not knowing there were others in the same town already on the team.  We are more focused on the engineering problem and its sequential steps than a deadline or being the winner.  We are also looking at what the production version will look like and how it will be operated in the field.

[JonHylands]

JonHylands,
Will your biped be something you will market and sell?

Eventually, although the first biped will be made from a Bioloid kit. The eventual one (which will take a few years to build) will look something like this:

http://www.huv.com/Robot-Raptor.jpg

The biped itself isn't all that important - lots of people have built those. What I'm going for is the holy grail of AI, which is really a software problem, although I believe you need a physical robot body to solve it.

Assuming I did eventually start building and selling the raptor robot, I would need to make an arrangement with Yobotics first, since they hold a patent on the series elastic actuator.

- Jon

[sdk32285]

militoy -
Which team r u with?

[Militoy]

Well - the announcement came out, and cash prizes for the race have been re-instated, but only for the top 3 finishers. 1st place gets $2 million, 2nd place $1 million, and 3rd takes home $500,000. It's a shame the track B teams will no longer get a cash kick, just for making the NQE, but some money for top finishers is far better than trophies!

sdk32285 - sorry I didn't see your post earlier - I'm on the "Chassis Team" on Team Scorpion. I'm in charge of the control interface between the Mission Processor and the Primitives on the vehicle, called the "PCU". It's kind of an unpaid side job for me, that's taking as many hours as my day job!