Roboticists from 47 teams are preparing to take part in Europe’s answer to the U.S. Department of Defense’s Grand Challenge (last year’s robotic car race aimed at encouraging research into autonomous cars).
This first European Land-Robot Trial, to take place in Germany on May 15, will pit against each other teams from nine countries, representing both academia and industry. Unlike the U.S. Grand Challenge, organized by the Defense Advanced Research Projects Agency (DARPA), which is a single 132-mile race in the desert, the European version will consist of three different events, putting robots to the test in urban, non-urban, and landmine detection and removal scenarios.
Despite the obvious comparisons with the Grand Challenge, the European organizers stress that their event is not so much a competition as an evaluation of existing technology. “The objective here is more an assessment of where the technology is today,” says Henrik Christensen, chairman of the European Robotics Network at Sweden’s Royal Institute of Technology in Stockholm and one of the event’s judges.
Indeed, whereas the Grand Challenge awarded a $2 million prize to the fastest autonomous vehicle to complete its 132-mile course, the European robotic challenge offers no cash incentives at all. Furthermore, instead of having one overall winner, the event will award commendations for different categories, including “best overall performance” and “most technical solution,” says Christensen.
The trials will take place in and around Hammelburg, a mockup of a town used by the German military for training exercises. In the non-urban course the robots will have to contend with a one-kilometer route containing ditches, barbed wire fences, cattle guards, fires, narrow underpasses, and inclines of up to 40 degrees.
The urban and landmine 500-meter trials will require the robots to negotiate doorways, stairs, partially collapsed buildings, and poor visibility from smoke or partial lighting. Along the way, they will also have to search for designated objects and report their findings back to base.
As if the trials weren’t enough, “military” obstacles capable of disabling a robot if struck will be placed randomly along the routes. Precisely what these objects are will not be revealed until the day of trials, says Markus Lueck, one of the event’s organizers. There’s a high degree of realism to the trials, he says, down to the type of fake explosive used in the landmines: “We use a special fluid normally used for training that is chemically nearly exactly the same as explosives.”
Besides the challenging nature of these complex tasks, the way in which contestants will be given their routes has also prompted some people to suggest that the European version will actually be more difficult than the U.S. Grand Challenge. In the United States, contestants are given a series of global positioning waypoints – like GPS breadcrumbs – leading them along the route. In the European event the only information teams will be given is an aerial image of the terrain with a route roughly sketched out.
This makes all the difference, says Roland Siegwart, whose team at the Ecole Polytechnique Federale de Lausanne in Switzerland is submitting an adapted Daimler-Chrysler Smart Car in the non-urban trial. In the Grand Challenge the task was more about travelling between GPS waypoints in as straight a line as possible, with some minor obstacle manoeuvres, he says. The European contestants have more freedom to plan their routes and use their sensors to map the environment for the safest or most efficient path.
But not everyone is convinced that the European version is harder. “The Grand Challenge was an amazing endurance test,” says Sebastian Thrun, who led the Stanford University team that won the Grand Challenge last October. In comparison, the European trials are extremely short, he notes, so that it is difficult to make a straight comparison.
The fact that European entrants do not have to be autonomous – semi-autonomous and even remotely operated vehicles can also be entered – may also lead some robot purists to favor the Grand Challenge. “From my point of view, a robot should be fully autonomous,” says Siegwart. But, for military purposes, robots often have a human in the loop, he adds, such is with unmanned aerial vehicles.
One way to settle this debate might have been to enter “Stanley,” the Stanford winner from last year’s Grand Challenge – but this isn’t going to happen. “I tried to participate but was denied access,” says Thrun. Being a native-born German now holding dual U.S. citizenship, in theory, Thrun should have been able to participate, since the rules only state that one must be a European citizen in order to qualify.
Meta has built a massive new language AI—and it’s giving it away for free
Facebook’s parent company is inviting researchers to pore over and pick apart the flaws in its version of GPT-3
The gene-edited pig heart given to a dying patient was infected with a pig virus
The first transplant of a genetically-modified pig heart into a human may have ended prematurely because of a well-known—and avoidable—risk.
Saudi Arabia plans to spend $1 billion a year discovering treatments to slow aging
The oil kingdom fears that its population is aging at an accelerated rate and hopes to test drugs to reverse the problem. First up might be the diabetes drug metformin.
Yann LeCun has a bold new vision for the future of AI
One of the godfathers of deep learning pulls together old ideas to sketch out a fresh path for AI, but raises as many questions as he answers.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.