The National Robotarium

Professor Rieser: Whether it's Google Home, Amazon Alexa, Siri or Bixby, many of us now have 'smart assistants' in our home but we often don't use them beyond setting kitchen timers.

The machines you've had experience with were most likely built using a vast set of complex rules -- hand-written by a large workforce of software engineers. In our research at Heriot-Watt, we're exploring a different model - and asking the question: Can machines "teach themselves" how to have a conversation?

We're looking at a method called 'machine learning' in order to teach machines to have a conversation just by looking at human examples meaning that machines could potentially learn to talk about anything from jointly solving a specific task to having a social chat.

This can be useful in many different contexts, from healthcare to entertainment and education.

Meaning that soon they could be for more than just weather forecasts and setting kitchen timers.

Dr Thusha Rajendran: Robots and artificial intelligence are a part of our future. So harnessing their powers to help people with autism all over the world today, will help millions tomorrow.

In the UK, only sixteen percent of adults with an autism spectrum disorder are in full-time employment but our research shows that with the right training, many more could fulfill their potential…helping them, their families and society.

So, how can robots help those with autism in the work place?

Our research at Heriot-Watt will help, through our pioneering use of a Flash Robot that can produce human-like emotions.

The aim is to help those with autism better understand the expressions of others, aiding them in the work place when interacting with co-workers and boosting their confidence at the same time.

The Flash Robot can also help employers understand how they may need to change their workplace practices to help employees with autism.

We want to ensure that today's technology can be used to make a difference tomorrow and by helping those who truly need it, our research can effect significant change.

Professor Lemon: OK Google… Hey Siri… Alexa - let's chat…

Voice interactive systems are all around us – at home, at work or even on your wrist.

At Heriot-Watt's Interaction Lab, we're developing technology that machines can understand to help them generate human language and join us in conversations.

Why are we doing this? Well, spoken conversation is the most efficient, natural, and enjoyable method of communication that we have. Allowing computers to speak with us naturally and easily will open up more devices and services to more people.

Our research is developing the technology that can make such human-level communication skills a reality - for future systems that will cooperate intelligently with us.

In fact, our AI system is currently leading the 2018 Amazon Alexa Challenge, an international contest to develop a socially intelligent conversational AI for the Amazon Echo platform.

Our Interaction Lab is one of the few places in the world that has pioneered the use of machine learning methods, meaning our research could help everyone – from people with disabilities to surgeons, drivers and cooks – to perform difficult tasks hands free, making even the most complicated things accessible and achievable in every-day life.

Professor Helen Hastie: We live in a world where robots are becoming more reliable and able to help humans in ways never previously thought possible.

In our research at Heriot-Watt, we're working on techniques that allow humans to work effectively in a team with robots and autonomous systems. We're researching how robots could be safely used in hazardous or difficult-to-reach environments, such as on offshore wind farms or deep underwater.

Through initiatives such as the ORCA Hub, we're developing techniques that allow the operator and robot to work towards a common goal, whilst at the same time keeping the human out of harm's way.

So why is this so difficult?

Well, the problem mostly comes down to lack of communication so we're looking at ways that the robot can explain what their doing and why, in natural language, and we think that this will enable a trusting relationship to evolve between the robot and the human.

Establishing these relationships will mean that we're more likely adopt robots to do these dangerous jobs and will result in less risk to life.