Where am I? Where am I going? And via what route? Those are the three key questions robots have to answer in order to move around autonomously within a given environment. In doing so, they employ a wide variety of different navigation solutions, which are frequently also combined, to ensure optimal orientation.
Robust autonomous navigation is essential to all mobile robot applications. For a mobile robot to be able to perform tasks such as “move to destination A” in different operating environments, it must be able to correctly identify its current location and to navigate. One of the basic methods used for this is global satellite navigation, as is also used in cars. The best known of them is without doubt the US Global Positioning System (GPS). Other countries have their own systems, too, including the Russian Glonass system, the emerging Chinese system Beidou, and Europe’s Galileo, which already has satellites in orbit. Satellite navigation offers positioning accuracy to within a few metres, depending on the number of satellites the given system can receive. As a result, robots are quite able to navigate through terrain, provided they have a map of the locality stored in their memory.
Satellite navigation with centimetre precision
To provide even more precise navigation, the signals from the satellites can be processed by software. US company Swift Navigation, for example, markets a software-based system which attains accuracy to within just a few centimetres using the standard positioning data from the satellites. Low-cost smartphone components provide the hardware. “This is not your cell phone’s GPS, which is accurate to about 10 feet. That’s good enough if you’re looking for a restaurant, but doesn’t come close to helping autonomous vehicles navigate the world,” said Tim Harris, CEO of Swift Navigation. “With our centimetre-accurate GPS, a car knows what lane it’s in, (…) and a drone can drop the package on your doorstep, not in your neighbour’s pool.” But satellite navigation does have one basic disadvantage: it stops working if the satellite signals are blocked, such as on a city street between skyscrapers, or inside a building. So a number of vendors have developed systems which replace the satellites with different signal sources.
Radio signals replacing satellites
For the indoor sector especially, Wi-Fi is a prime candidate: Wi-Fi routers, whose position and signal strength is known to the system, can be used to locate the position. In very densely populated areas, this enables a position to be located with an accuracy of 10 to 20 metres in just a few seconds. WPS (Wi-Fi Positioning System) is usually used in conjunction with other systems, such as GPS. Combining these systems creates a hybrid form of navigation which significantly improves the performance of the overall navigation system. Similar systems are also realised using Bluetooth technology.
A prime example in this context is Apple’s iBeacon technology. It involves placing small transmitters at predetermined points in a building, which are marked on a digital map. The transmitters send signals at fixed time intervals with unique identifiers. The data transfer runs via Bluetooth Low Energy technology, which means the transmitters can also be battery-operated. The receiving robot detects the transmitter’s identifier, measures its signal strength, and compares the data against the digital map. If it is receiving from four transmitters, it can even determine its location in three-dimensional space.
A still relatively new technology is Ultra Wideband (UWB): It involves installing transmitters which emit wideband signals of more than 500 MHz in the frequency range between 3.1 GHz and 10.6 GHz. If the robot knows the positions of the transmitters, it can determine its position by triangulation. The system devised by Irish company Decawave, for example, guarantees an accuracy of 10 centimetres on the basis of UWB technology. “The market for next generation indoor location technologies with improved accuracy is beginning to advance with solid use cases and adoption. UWB is clearly carving out its space, with ABI Research forecasting strong growth across a range of verticals,” said Patrick Connolly, Principal Analyst at ABI Research. “The market opportunity is quite large and companies like Decawave that are leading the charge in UWB are well positioned to experience continued growth.”
Navigation based on natural features
An alternative – or addition – to navigation by radio signals is offered by onboard systems which map the surrounding environment. They use radar or lidar. Lidar (Light Detection and Ranging) is a technique in which a pulse of light is emitted and a distance can be calculated based on the propagation time and speed of the light. Lidar is closely related to radar as a method of optical distance and speed measurement, though using laser pulses rather than the radio waves used by radar.
Ultrasound systems operate according to the same principle. They utilise special chips which emit sound waves in the ultrasonic range. If the waves encounter objects, they are reflected back and received again by the chip. The distance to the object can then be calculated from the propagation time of the sound wave.
Another method of 3D mapping is by stereo cameras: Colour and depth cameras generate a pixel cloud with exactly assigned distance values. On that basis, and by comparing against a previously created map, the robot is able to plot its position very accurately. Laser scanners are a solution for navigation typically employed in areas such as large-scale warehouses or in automotive applications. They are able to identify reflective targets in a 360 degree range by means of a rotating laser beam. A robot can determine its position by calculating the distance and angle to the target – provided it is able to cross-check the reference points against a previously created map of the surroundings. “This approach is today referred to as ‘natural feature navigation’,” says Nicola Tomatis, CEO, Bluebotics. The ANT (Autonomous Navigation Technology) system developed by the Swiss company achieves an accuracy to within one centimetre using conventional safety laser scanners for navigation. To do so, however, the system combines the data from the laser scanner with measurement values from additional sensors which monitor the movement of the robot. These may be sensors which monitor the rotary motion of the drive wheels, and so measure the distance travelled, or gyrometers which measure the robot’s rotary orientation.
Navigating by light
Robot manufacturer Adept has enhanced its indoor navigation systems with so-called “Overhead Static Cues” technology: If the surroundings have been altered so much – such as due to pallets or boxes lying around in a warehouse – that the measurement results fall below a certain probability of recognition, an upward facing camera can be optionally used to supply additional sensor data. The camera is oriented to the ceiling lighting for example. Based on a minimum of three visible ceiling lights, the position in the room can be additionally determined and the other sensory data corrected accordingly.
Researchers at Lund University in Sweden are also focusing on light. They have devised a concept for a new drone orientation system based on the evasion techniques of bees: The insects assess the intensity of light penetrating through holes in the dense undergrowth in order to evade objects. The researchers claim the system can be ideally tailored to small, lightweight robots. “I predict that our vision system for drones will become a reality in five to 10 years,” says Emily Baird from Lund University. As she points out, using light to manoeuvre around complex environments is a universal strategy which can be deployed by both animals and machines. Detecting and safely passing through openings is made possible as a result. “It is fascinating that insects have such simple strategies for solving difficult problems for which scientists have as yet not found solutions,” Emily Baird concludes.
(Picture Credits: Unsplash: NASA, SpaceX)
