Real robots have captured the imagination of young and old engineers, designers, and programmers alike. For example, techies can get their hands dirty with iRobot’s Roomba Create or take part in competitions like FIRST Robotics. Robots also are changing how war is waged and how we protect people on and off the battlefield. And, they’re working with doctors and patients. Though we’re far from the intelligent androids found in science fiction, robot deployment and purchases are on the rise.
Still, humanoid-like robots continue to drive research and development. One such project has kept graduate students at Mälardalen University in Sweden on the run to keep up with their latest creation, nicknamed Dasher (Fig. 1). Led by Prof. Lars Asplund, a team of 21 students created Dasher using a combination of electronics, hydraulics, and an exoskeleton made of titanium tubes. Overall, it’s comparable to a human in size and stride.
Dasher is designed to run so it forgoes features such as hands, though it has arms for balance. It employs the prosthetic lower leg technology utilized by Paralympics runner Oscar Pistorius. This technology was supplied by the Icelandic company Össur, which also equipped Pistorius with his prosthetic limbs.
These design choices reduced the complexity of an already ambitious project while retaining the basic charge of developing a robot that could run using its own locomotion and balance. The team’s choice of Adacore’s GNAT Ada tool suite as the programming system was critical to delivering a workable system as well.
Ada’s Ravenscar tasking profile, a subset of the Ada language, allowed Dasher to meet the real-time requirements. Automated development tools were created to transform the Ada code into timed automata that were then analyzed by the formal verifier UPPAAL.
The Ada applications are hosted on a centralized Freescale PowerPC card that runs the Wind River Systems VxWorks real-time operating system (RTOS). Each arm and leg has its own 8-bit Atmel AT90CAN128 controller-area-network (CAN) microcontrollers that connect to the PowerPC via a CAN bus. The smaller micros manage the hydraulic servos and the dc motors, and the larger micros drive the hydraulic compressor.
The dual-camera vision system and three-axis gyroscope drive a Xilinx XC2V8000 Virtex II FPGA, which handles real-time image and distance recognition to recognize the two white lines on a running track. The goal is to have Dasher run, not walk, on its own around a track. It should finish a 100-yard dash in 9.5 seconds.
ROBOTIC STAND-INS
HRP-4C sounds like a partner for C-3PO (Fig. 2). But this robot is a more lifelike creation from the Humanoid Research Group, which is part of the Intelligent Systems Research Institute of the National Institute of Advanced Industrial Science and Technology (AIST). It stands 158 cm tall and weighs 43 kg, including the battery. The physical dimensions and structure were based on an average young Japanese female. HRP-4C was developed as part of the User Centered Robot Open Architecture (UCROA).
Its movement and walking motion, based on motion-capture, is designed to mimic a person’s gait and gestures. It also includes speech recognition and voice synthesis. The robot is intended for applications in the entertainment industry, including use at fashion shows. In fact, it appeared at one held during the Eighth Japan Fashion Week in Tokyo this year.
WowWee’s feminine Femisapien (cover image) and hunky Joebot (Fig. 3) are smaller than HRP-4C. They also cost less than $100, while the HRP-4C specs out around $200,000. Joebot is the newer and more capable of the pair, but both walk and talk. They have different levels of voice recognition, with Joebot responding to key phrases. Joebot uses IR sensors for obstacle recognition. Each robot has limited programmability, too.
WowWee also has a telepresence robot called Rovio (Fig. 4). This Wi-Fi-enabled robot can be controlled by most Web-enabled devices, from a PC to a cell phone. It even has Universal Plug and Play (UPnP) support. The head-mounted movable camera has a resolution of 640 by 480. There’s even a new LED headlight option for operation in dimly lit areas.
The Rovio is very mobile with its omnidirectional wheels. A TrueTrack Navigation System enables it to locate objects within its environment. IR sensors are used for obstacle avoidance. Rovio has its own docking station and can automatically find its way home to recharge its nickel metal-hydride (NiMH) batteries. The robot is programmable, allowing it to follow pre-programmed paths, and can be remote controlled. The Web-based application programming interface (API) is available for download.
Anybot’s Anybot QA telepresence robot is more like a cross between Venus de Milo and Disney’s Wall-E (Fig. 5). It was seen wandering around this year’s Consumer Electronics Show (see “See CES From Another Point Of View”). The Anybot QA isn’t articulated. Rather, it balances atop a pair of wheels, providing an efficient mobile platform. Balancing doesn’t take a lot of power, and turning in tight places with a pair of wheels is almost trivial. The Anybot QA strolls along at up to 6 mph.
This robot uses a Wi-Fi or 3G cellular connection to exchange audio and video with its controller. A pair of 5-Mpixel cameras captures local video while a 7-in. LCD displays video at the controller’s end. Forward-looking light detection and ranging (LIDAR) is used for obstacle recognition. A green laser pointer eliminates the need for hands and arms to point at objects.
The 30-lb robot runs off a set of lithium-ion (Li-ion) batteries. Runtime is four to six hours. The control software runs on a PC or Mac with audio and webcam support. Also, the Anybot QA stands about 5 ft tall and has just two joints. One is for the head, which tilts forward, and the other is in the lower middle for balance.