The IET Robotics and Mechatronics TPN was delighted to exhibit at the European Robotics Forum 2017 in Edinburgh last month. The conference took place over the course of three days and brought together end users, technology developers, service providers and academics from a cross section of industry, for three days of presentations and panel discussions.
It was fantastic to be surrounded by the biggest and best innovators in European robotics and there was a strong sense of collaborative thinking and knowledge sharing.
Many themes were discussed at ERF but we've picked 3 which we found most interesting:
It was fantastic to be surrounded by the biggest and best innovators in European robotics and there was a strong sense of collaborative thinking and knowledge sharing.
Many themes were discussed at ERF but we've picked 3 which we found most interesting:
- Focus on Artficial Intelligence
It seems that we have now reached a state where we (and potentially society) have accepted the fact that the capability for robots to ‘think’ and act with superhuman powers is within our lifetime’s reach. This is more so given the recent successes in AI machines (AlphaGo beating the best in human expertise). Google DeepMind is making massive strides in this field with successes growing at an exponential rate. Regulatory frameworks are being set up and the robotics community needs to look into providing safe ‘thinking’ robots to assist humans with complex problem solving and not just physical power augmentation.
- Robotics as a service
There was a lot of discussion on the idea of robotics as a service rather than being sold as a product, and there are already some signs of mobile robotic platforms becoming commoditised, with capability added through bolt-ons or software apps. This ensures manufacturers are focussed on developing open interfaces to the end user, which could be a commercial user with a specific application or tool requirement, or a researcher looking for rapid deployment of a leading edge sensor or behavioural development.
- Neural Network architecture
Various Neural Network architectures are being redesigned, augmented or reconfigured and also, given the augmented computing power and cloud computing services available, large networks can be developed and huge amounts of data and iterations can be handled. This however might still pose a challenge on embedded devices. The idea of having efficient training mechanisms still exist especially on real robots and embedded devices. An alternative to this is training in simulation and then transferring the knowledge onto the real machine. This tends to be hard to do in practice and need to be looked into in much more detail. This is expected to be an emerging field of research.
Did you attend ERF this year? What was one topic discussed that you found of particular interest?
Many thanks to Raphael Grech and Lee Wilson for contributing to this post.