robotics

Dr Nick Hawes will talk to the AI and Robotics Innovation forum today (Thursday 13 March) about the use of robotics and artificial intelligence in near-future commercial applications. 

Watch a video interview with Dr Hawes online here.

According to JD Supra, robotics is the fastest industry in the world. The market for industrial robots, i.e. reprogrammable machines which can do fixed, predictable tasks, has started to grow again after a few years of stagnation. A greater growth is expected for service robots, which are robots which have a degree of *autonomy* -- the ability to makes decisions for themselves. It is this autonomy which provides the potential for growth: when machines can be augmented with sensors and reasoning capabilities they can solve problems that can't fully specified in advance, and this opens up a much greater range of application areas. Such applications could be ones where sensors are required to understand what a robot needs to do (e.g. locating litter to pick up) or where reasoning needs to be done to work out the actions to perform (e.g. which route to take through the rubble in a disaster area).

The fields of robotics and artificial intelligence (i.e. the techniques necessary to create artefacts which sense, plan and act) have long held promise, but have seldom delivered results to match this promise. This is now changing, with companies such as Apple, Amazon and Google investing heavily in robotics-related technology. This new growth has been caused by a number of factors. The first is the availability of cheaper, more reliable sensors, particularly sensors capable of providing the 3D images necessary to create and understand models of the *unstructured* environments service robots are expected to work in (institutions and offices rather than factories). The need for this kind of sensing in household technology is behind both Apple's purchase of Kinect originator PrimeSense, and Dyson's £5 million collaboration with Imperial College. 

Given this sensor information, robots must be able to work out what to do from it. The second generator of recent growth is the ever-growing amount of open source robotics software which can be used to do everything from building 3D maps and planning driving routes, to making decision in uncertain situations and interpreting speech. Having this software means it is now relatively easy to build an autonomous robot containing state-of-the-art algorithms. The power of the availability of such software is clear to see in Google's recently announced Project Tango which builds on the results of many open source projects.

Finally, these developments are being driven by large-scale research programmes which bring together academia and industry to continually evolve the understanding of the basic understanding of robotics and AI, and exploit it in novel applications. The Horizon 2020 robotics programme in the EU, and the DARPA Robotics Challenge and Roadmap for Robotics in the US, will see a new wave of intelligent, autonomous systems making a difference to our lives.

Given all this progress, where should businesses be looking to introduce service robots? Despite the characteristic worry that robots will come along and take jobs from humans, it is actually hard to imagine this happening. Humans are amazing at a huge range of things; robots are good at a very small range of things, but what they can do, they can do predictably and indefinitely (but often slower than a human can). Given this, it makes sense for companies to look to add value to their services with robots working *alongside* humans, rather than replacing human wholesale. The introduction of robots should start initially with tasks that robots can currently do well, which is mostly safely moving around indoors and detecting a range of of stimuli (from QR tags to faces). It will also naturally happen in situations where the cost/benefit trade-off of investment in robotics technology is positive. This is why industrial robots first appeared in large-scale manufacturing (huge scale and value, highly controlled environments) and why one of the first breakthrough applications of service robotics was Amazon's shelf-carrying Kiva warehouse robots. We're working towards similar breakthroughs in the security and care industries with our EU-funded STRANDS project which aims to create long-running mobile robots capable of learning about their environments on a day-to-day basis. 

This development trajectory naturally rules out the sci-fi dream of personal robots doing household chores for us as the cost will be prohibitive for most individuals (in the short-term), and the technology for such tasks (notably robot manipulation in unpredictable environments) is not yet ready for commercial deployment. However, wider industrial uptake of autonomous robots will help drive development for more niche or challenging applications (e.g. assisted living for the elderly), and may one day lead to a robot in every home.

This article was first published on City AM