It has been about four decades since a human set foot on the moon. With space agencies facing major financial challenges, missions to send people back to the moon and elsewhere in space are on the back burner. But that isn't stopping IEEE members from trying to expand our knowledge of the universe with nonhuman explorers: robots.
Space robots come in many shapes and sizes. Planetary rovers explore the surface of moons and planets, taking photos and soil samples and sending the data back to labs on Earth. Orbital robots, a relatively new type, service orbiting satellites or assemble parts of a space structure. The Japan Aerospace Exploration Agency (JAXA), the U.S. Defense Advanced Research Projects Agency, and other organizations are experimenting with orbiting prototypes. And probes are landing on asteroids. JAXA's Hayabusa probe, for example, alighted on the surface of the asteroid Itokawa in 2005 to collect soil samples. Hayabusa is expected to return to Earth this month.
The first robot to explore an extraterrestrial body was Lunokhod, launched in 1970 by the Soviet Union to explore the moon. Remotely operated from Earth, Lunokhod traversed 10.5 kilometers of the moon surface, taking photos and analyzing the soil.
During the past decade, there were major breakthroughs in space robotics, notably by NASA's autonomous Mars rovers, Spirit and Opportunity, which landed on the Red Planet in 2004 and continue to send back photos and data.
The two rovers have encountered numerous obstacles. Opportunity got stuck in the soil for several weeks in 2005 before engineers could get it moving again. And Spirit is now immobile, trapped in sand, struggling to tilt its solar panels toward the sun for some extra electric heat before the extremely cold Martian winter hits.
The rovers' troubles have helped researchers such as IEEE Member Kazuya Yoshida understand how to build a better crop of robotic explorers. Yoshida, a professor of aerospace engineering at Tohoku University, in Sendai, Japan, is an expert on the mechanisms and control of space robots, including planetary rovers and asteroid robots such as Hayabusa, which he helped design. He was featured in the December 2009 issue of IEEE Robotics & Automation Magazine, which was dedicated to terrain mapping, sensors, robotic-wheel traction control, and other space robotics technologies.
"For lunar and planetary exploration, robots are especially critical, because human access to the hostile space environment is very difficult or in many cases not yet possible," Yoshida says.
INTO THE UNKNOWN
Yoshida and his team of researchers at the university's Space Robotics Lab have been working on several advances in planetary rovers. One involves improving the topographic mapping techniques robots use as they prepare to traverse the landscape. Robots map their environment using sensors and then plan a safe route. The researchers are also working on modifying the traction mechanics of the robot wheels to prevent rovers from getting stuck in soil.
The lab's most recent creation, a robot named El Dorado II, showcased its advances at the planetary rover contest during last year's IEEE Robotics and Automation Conference, in Kobe, Japan. Five robots tried to move about a Mars-like terrain. The fully autonomous El Dorado II, which snagged first place, was the only one to successfully map the test course, navigate across the bumpy gravel field littered with rocks, reach a target, and return to its base. The robots' human minders had no prior knowledge of the course, so their robots had to deal with whatever they encountered, including rocks and sand.
To make sure El Dorado II can overcome obstacles, Yoshida has been working with new sensing and navigation techniques. To map its environment before it sets out, El Dorado II uses a cutting-edge technique called 3-D simultaneous localization and mapping. The robot uses laser sensors to capture layers of its surroundings by measuring the dimensions of the topography within a range of about 30 meters. It builds a three-dimensional map by superimposing the multiple layers to form a complete view of the area.
El Dorado II uses an iterative closest—point algorithm to construct the map. Originally, the algorithm was developed as a way of registering 3-D shapes for computer graphics, but it is now widely used in mobile robotics to construct topographic maps. Once the map is completed, El Dorado II determines the best path for avoiding obstacles.
As the Mars rovers demonstrated, it's pretty easy to get stuck but difficult to get free. Yoshida is working on ways to ensure a robot doesn't get trapped in the first place. That means focusing on the wheels.
NASA's Mars rovers are successful at moving over bumpy, rocky terrain, he notes, but they have a more difficult time going over sand, because the wheels can slip, causing them to grind into the ground and get stuck.
Yoshida and his team developed a system that addresses the wheels' slip ratio, which is the difference between a wheel's tangential speed and the speed of the axle relative to the ground. When the wheels have different slip ratios and one wheel is going faster than the others, the robot can get stuck. "The idea," Yoshida says, "is to control the rotational velocity of each wheel so their slip ratios are equal."
To do that, El Dorado II uses an onboard video camera to observe the terrain texture beneath it. By differentiating the successive terrain images, optical flow vectors are obtained. Sensors in the robot's inertial measurement unit analyze the vectors to estimate the slip ratio and slip angle of each wheel. Based on that, the robot's control system adjusts the speed and angle of the wheels to minimize slippage.
Yoshida says he hopes his space-robotics work helps society explore parts of the universe impossible for humans to visit. "Robotic exploration is absolutely necessary to expand the horizons of our knowledge and presence in space," he says. "Robots should be a precursor to any future human expeditions."