“If autonomous control is ever to equal or surpass the ability of humans, the ability for the technology to ‘see’ better than humans in adverse conditions is imperative.”
While early predictions that fully-autonomous vehicles are just around the corner have now given way to reality, the relentless march toward that ultimate goal continues on multiple fronts. Bit by bit, progress is being made. Reports of new on-road trials and new players in the game continue to arrive in a near-constant stream.
One big issue that hasn’t yet been fully addressed or resolved, however, is the one that may be most important to us Canadians: how will autonomous vehicles cope with winter?
It’s one thing for AVs to perform satisfactorily in the benign climates of Arizona or California, which are preferred areas for many feasibility demonstrations. It’s quite another to consistently and reliably get the job done at extreme sub-zero temperatures, in blizzard conditions. Or even in more moderate conditions, when visibility is poor and road markings and other indicators are covered with snow, ice and dirt, which many of us face every day. Or conditions like the blowing-snow whiteouts like that contributed to a recent 200-vehicle pile-up in Montreal.
It’s a question that is important in other parts of the world, too. Places like Russia.
For that reason, three of that country’s state development institutions (the Skolkovo Foundation, RVC and the Agency for Strategic Initiatives) joined forces to establish a new format of technology competitions collectively known as ‘Up Great’, to help stimulate autonomous technology development by Russian industry and academia.
That series of competitions is unabashedly patterned after the DARPA Grand Challenge of 15 years ago, a competition for autonomous vehicles funded by the U.S. Department of Defense, which is now widely regarded as the launchpad for development of AV technology in America.
A team from Stanford University won the $2-million prize offered for completing a 212-km course across the desert in the shortest time (6h:53m;58s). And many of the members of that winning team went on to form the core of Google’s AV development team (now Waymo).
Reports of new on-road trials and new players in the game continue to arrive in a near-constant stream.
The first of the Russian challenges, called “Winter City” was somewhat simpler in that its goal was for prototype AVs to travel 50 prescribed km in less than three hours. But rather than on open desert, it had to do so on a closed testing course simulating city streets, complete with pedestrians, strollers, bikes, and other forms of transport, as well as construction zones and barriers and stunt drivers running red lights at intersections. Plus they had to do so in the depths of Moscow’s winter, supplemented by snow blowers creating blizzard conditions, hiding road markings and signs beneath a thick covering of snow.
The federal state contributed 162-million rubles ($3.25M Cdn) to organize the ‘Winter City’ contest among AV developers and another 175-million ($3.5M Cdn) to the winner.
While 33 teams initially applied for the contest, only nine made it to the first round of qualifying last February and just six of them successfully qualified for the competition held in December. Five actually took part in that final round.
And, while there was a winner — the entry of the Russian firm StarLine — no team met the criteria necessary to win the $3.5M (Cdn) cash prize. The winning StarLine entry satisfactorily completed the 50km course but took four hours to do so.
Admittedly, it is unlikely that any of the Russian entries were as fully developed as the latest prototypes from organizations such as Waymo, GM’s Cruise, Ford/Argo or the like. But the exercise also highlighted the fact that winter remains a major challenge to be overcome.
Among the specific factors identified by participants as important in winter use was “the ability of a driverless car to operate in adverse weather conditions, in the absence of visible markings and with high levels of interference (from other traffic) and other unforeseen hazards.” Exactly the kind of conditions Canadians routinely experience throughout a typical winter.
Not surprisingly, given the conditions in the Russian competition, vehicles equipped with Lidar technology (discussed previously in this column) appeared to have an advantage over those with only radar and camera systems. It seems a reasonable conclusion that if autonomous control is ever to equal or surpass the ability of humans, the ability for the technology to “see” better than humans in adverse conditions is imperative.
Whether that ability is ultimately achieved by Lidar or thermal imaging or some other combination of technologies, it remains the holy grail of AV development, at least in terms of potential winter use. Might a Canadian parallel to Russia’s ‘Winter City’ challenge help Canadian industry, academia and government achieve it?
