Robotics Tech Digest - May 2017

Colonoscopy robot

Researchers at Vanderbilt University in the US have developed a robot that can be used to perform a colonoscopy. The semi-autonomous system consists of an 18-mm capsule robot that is guided by magnets outside the body as it examines the colon. A magnet attached to a robotic arm can then be used to guide the robot through the colon which promises to be more comfortable for patients than having an endoscope pushed through. The system has been tested on a pig that underwent 30 magnet guided colonoscopies, the researchers hope to begin human trials in 2018. 

Transferring information from robot to robot

Researchers at MIT’s CSAIL (Computer Science and Artificial Intelligence Laboratory) have developed a system that bridges two techniques commonly used to program robots: learning from demonstration (see and repeat), and motion planning (coded instructions).  The researchers’ combined method is called C-LEARN. Its simplicity will allow noncoders to teach robots a range of tasks by providing basic information on object manipulation and then demonstrating the task to the robot. Perhaps most importantly this method allows for skills to be transferred to other robots that have different ways of moving. The researchers tested the system on Optimus, a two-armed bomb disposal robot, that they programmed to carry out a variety of tasks. The information from Optimus was then transferred to Atlas, a humanoid robot, that manage to assimilate the information and carry out the tasks as well as Optimus. The flexibility that this system gives a robot – not being programmed to carry out an operation in only one way – means that robots would have far more autonomy and versatility. 

Robots in security

Dubai, UAE, has inaugurated its first robotic policeman, Robocop. Robocop will be stationed in malls and tourist attractions allowing people to tap on its touchscreen chest to report a crime, pay traffic fines or have a ‘chat in English or Arabic’. The robot’s inbuilt camera will livestream video to the police command centre.  The Dubai police are hoping that future versions will have facial recognition software and an enlarged set of languages. Dubai is aiming for 25% of its police force to be robotic by 2030 with plans for driverless patrol cars and scooters, egg-shaped parking warden robots, and a three-metre-tall robot capable of running at 80kph. 

Robot controls movement of turtle

The Korea Advanced Institute of Science and Technology (KAIST) has developed a parasitic robot system that it has attached to a turtle. The robot induces the turtle to act in a way that benefits the robotic system. In this experiment the robot was attached to the top of the turtle’s shell along with a feeding bottle and a bow of LEDs that reach forward in front of the turtle’s head. The turtle is trained to go to certain waypoints in a tank of water with a LED lighting up in the array indicating to the turtle which direction it should move. When the turtle moves in the correct direction it is rewarded with food. Over time the behaviour of following the LEDs’ indications is reinforced and becomes a learnt behaviour. The scientists see this as having possible uses in exploration and reconnaissance missions which a robot or human would find difficult alone. 

Soft robot legs

University of California San Diego engineers have developed soft robotic legs that can walk on rough surfaces like pebbles or sand, and climb over objects. The 3D printed legs are composed of three parallel, connected sealed pneumatic actuators made of a rubber-like material. The relative degrees of inflation, and choice of which of the actuators is inflated, give a high level of control over the movement of the legs. The current four-legged prototype robot is tethered to an air pump and board-based processing unit, although the scientists are investigating ways to miniaturise these so that they could be included in the robot itself. 

Oil and gas rig inspection robot

Taurob, a German robotics company, together with TU Darmstadt, has been announced the winner of Total SA’s ARGOS (Autonomous Robot for Gas and Oil Sites) Challenge. Total, an oil and gas company, was looking for automated inspection robots that could be used to inspect and maintain oil and gas rigs. The winning robot can read pointed instruments, fill-level displays, and valve positions using its camera and scanners. It is further able to monitor temperatures and gas concentrations, detect abnormal noises, obstacles and nearby people, as well as safely navigate wet stairs. Taurob says that the robot is the first fully automated robot that can be safely used in explosive environments. Total plans to have the robot on its oil rigs from 2020. 

Optical bot

Scientists at Tampere University of Technology (TUT) have demonstrated an optically actuated robotic gripper. The gripper – an elastomer strip about one centimetre in length – acts like a Venus flytrap plant through manipulation with light from liquid crystal elastomers. The strip is glued to optical fibre through which blue light is transmitted. When an object enters the light field being emitted by the optical cable, indicated by light reflection, the elastomer closes up and coils around the object. The use of reflectivity would allow the device to only trap certain specified objects if needed. The lifting power of the device is hundreds of times its own body weight. 

Robotic lettuce harvester

Taylor Farms, California, is using an automated romaine lettuce harvester that uses water jets to cut the head from the root. The robotic harvester is attached like a large sidecar to a tractor and driven across the harvestable field. A human stands atop the harvester to inspect the picked heads for quality. The farm says that using the harvester improved quality and yield of lettuce compared to traditional harvesting methods. 

Robots in the airport

SITA, a company specialising in air transport technology solutions, has announced the rollout of two of its robot solutions in airports in Geneva, Switzerland, and Buenos Aires, Argentina. At Geneva Airport the company is using its kiosks – that with the exception of being mobile and autonomous work in the same way as current fixed self-check-in kiosks –  to go to busy and congested areas of the airport as needed to help reduce check-in queues. 
Leo is being used in Buenos Aires Airport to help passengers not only check-in but also print bag tags and take their bags away for processing. The passenger can insert their baggage into a secure box at the base of the autonomous, mobile check-in desk. Leo will then take the bags to where they need to go. 

Robot lands a 737

Aurora Flight Sciences with funding from DARPA (Defense Advanced Research Projects Agency) has demonstrated a robotic arm system, named ALIAS (Aircrew Labor In-Cockpit Automation System), landing a Boeing 737 in a flight-simulation program. Aurora posted a video on YouTube showing the robotic arm using the aeroplane’s control panel’s knobs and buttons to reduce the plane’s speed, pull or push levers to adjust the tilt of flaps, set the autobrake and landing gears to on, and pulling reverse thrusters as the plane touches down. Aurora says that the robot could work in cooperation with a human pilot to increase the efficiency of current flight operations, reduce pilot workload and improve pilot performance. 

Add this: