In this new series of articles, we will be exploring the past successes and failures, the present state of the art and the future potential in the automation of driving in logistics. We will look at the technologies, organisations and people driving the momentum in the autonomous vehicle development.
It was a fair October morning in a desert valley near Barstow, California. The sun had just gone up. Its light was gently warming up the air and glittered over a number of peculiar vehicles surrounded by a moderately sized crowd. What brought all these people up here so early, just past 6 o’clock in the morning?
There are commercial and university teams, Department of Defence representatives and camera crews present. The attention is focused on 23 participants: an eclectic collection of SUVs, trucks, dune buggies, ATVs and other full-wheel drive vehicles. There are but two common traits among this diverse starting field: a forest of sensors covering every square centimetre on each vehicle and no one manning the steering wheels.
Welcome to DARPA Grand Challenge 2005.
First Time Unlucky
In 2002 Defense Advanced Research Projects Agency (DARPA), a unit responsible for sourcing and developing emerging technologies for the needs of US Armed Forces, issued a challenge to researchers to develop a self-driving vehicle and complete an off-road race exceeding 240 kilometres in the rugged conditions of Mojave Desert on the border of California and Nevada.
DARPA’s aim in sponsoring the challenge was to encourage the development of a logistics solution for the military. The end goal is to automate the resupply convoys and take the drivers and other military logistics personnel away from the dangers of conflict zones.
Two years later, in 2004, a lineup of 15 challengers emerged: teams from Berkley, Caltech, Carnegie Mellon University as well as a number corporate and even high school teams all fielded their vehicles to race. All 15 failed to finish, one overturning at the start, some experiencing high speed crashes in qualification rounds, others plagued by software and hardware failures or just simply by a flat tire. Carnegie’s Red Team HUMMER vehicle named Sandstone went the farthest, getting stuck off course at 11.78 kilometre mark.
The Grand Breakthrough
Unfettered by the failures, many of the teams returned the next year to face 212 km long and complex route. This time, they were joined by the likes of Stanford, Cornell, Princeton, Virginia Tech Universities and UCLA.
Per the Challenge rules, the teams were provided with the final route details and GPS waypoints only on the night before the start of the race. The vehicles had to navigate and finish the course by themselves. Absolutely no manual intervention was allowed. There were more software failures, crashes and other mishaps this time around, but in the end, five winners emerged: Stanford, two Carnegie Mellon’s HUMMERs, Gray Insurance / Tulane University’s and Oshkosh Terramax vehicles.
Stanford’s Stanley vehicle based on VW Touareg was first over the line and completed the course in 6 hours 54 minutes at an average speed of just over 30 km/h. But an equally eventful and significant result came at the credit of truck manufacturer Oshkosh’s team and their 6×6 Terramax truck. Although it had to spend a night idling in the desert, Terramax successfully completed the route, including the challenging mountainous section at Beer Bottle Pass, in 12 hours 51 minutes. As the winners, Stanford Racing Team took a $2 Million prize home.
Nexus of Autonomous Driving
Why is DARPA 2005 Challenge held in such high regard in autonomous vehicle development history? There were self-driving car projects before that point, however, the Challenge brought the best and brightest minds together and drew public attention to their efforts. It was also followed by DARPA Urban Challenge in 2007 which tested autonomous vehicles on the roads of a closed Air Force base in California. Red Whittaker of Carnegie’s team led Tartan Racing to the first place in urban environment trial.
By and large, the participants of DARPA challenges doubled down on their efforts to improve autonomous vehicle technology.
Stanford team leader Sebastian Thrun went on to found Google X Lab and worked as company’s VP on development of the self-driving car project which spun off in 2016 as Waymo, a stand-alone subsidiary trialling autonomous taxi service in California, Arizona and Texas.
Chris Urmson of Carnegie Mellon’s Red Team also worked at Google as the CTO of a self-driving project and co-founded an autonomous car technology startup, Aurora in 2016 with his university colleague and teammate Drew Bagnell.
James McBride who helped stage the 2004 Challenge and participated in 2005 and 2007 founded the Autonomous Vehicle program and rose up to be Senior Technical Leader of Robotics and AI at Ford.
David Hall of Team DAD (Digital Auto Drive) and founder of Velodyne, a leading manufacturer of laser imaging and distance measuring sensors (LIDAR), used DARPA Grand Challenge as an opportunity to apply the technology in automotive solutions and improve on it. Hall and his brother supplied LIDARs to five of the six teams that finished the 2007 DARPA urban race. Nowadays, a large majority of autonomous vehicles under development either employ Velodyne LIDAR packages or have used them in the past. The company so far has sold over 30,000 sensor packages at the cumulative revenue $500 million since the sales started in 2007.
Defence and truck manufacturer Oshkosh who participated in the 2004, 2005 and 2007 Challenges received funding from the Department of Defence to develop their autonomous driving technology for operational deployment. Most recently in 2018, it was awarded a $49 million contract to integrate existing vehicles with scalable autonomous technology as part of the U.S. Army’s Expedient Leader-Follower program. Under the contract, Oshkosh will produce up to 150 autonomous kits for the existing 10×10 heavy tactical PLS trucks.
Only the Beginning
Following the needs of the US Army, DARPA Challenges laid the groundwork for platooning technology program run by Ground Vehicle Systems Center TARDEC. This technology allows an entire convoy of resupply vehicles to be operated by a single human driver with other trucks automated and following the lead. The concept has since been developed for commercial applications as well.
However, the impact of the 2004, 2005 and 2007 DARPA Challenges is far greater. They defined some of the core concepts of autonomous driving technology, captured public imagination and helped to fund the projects of the best and brightest in the field. To this day, wherever you go to autonomous vehicle manufacturers and self-driving technology suppliers – you will find familiar faces from that remarkable race in the desert.
Photo: DARPA / Oshkosh, 2005