Dexterous robots: bringing science fiction to life
27 August 2024
This blog has been co-authored by Prof Jenny Read, Programme Director for ARIA’s Robot Dexterity programme and Olivier Usher, Director at Challenge Works for prize research and design.
Are robots amazing or are they underwhelming?
Picture a future where robots work alongside us. What do you see?
We imagine a future where robots are collaborative, lending a helping hand in every task possible: sorting recycling, operating machinery, and preparing lab samples. The opportunities are endless, but this future still seems a long way off. Why aren’t robots reaching their full potential?
Take a look at YouTube and you’ll quickly find impressive videos of robot dogs navigating rough terrain, robot arms playing ping-pong or robotic humanoids doing somersaults. But look carefully and these demonstrations aren’t quite what they seem.
The environments they’re in are often highly constrained, sometimes they’re being operated remotely by humans, or the tasks they’re doing don’t require much dexterity. What’s more, they often fail many times on a task before getting it right – and you only see a clip showing the successful attempts.
Clearly, robots struggle with basic tasks, particularly when they’re doing things they weren’t specifically designed to do. They can be jerky. They can fumble. They’re often surprisingly imprecise. We have much progress to make in order to reach the future we envisage.
We can enable new advances in surgical techniques, we can make extreme environments safer for humans, and we can open new frontiers in space exploration, but we need greater innovation to get there.
So what can we do about it?
These are questions Challenge Works and the Advanced Research and Invention Agency (ARIA) are exploring, with potential to launch a challenge prize next year.
Strategies for advancing robotics
Companies and labs are investing vast resources in making robots smarter. Thanks to advancements in computing and AI, robots can see, sense and interpret their surroundings far better than ever before – and that means they can interact with the world in more sophisticated and precise ways. Most of the advances in robot capabilities in recent years have come from this approach.
But software is only part of the puzzle when it comes to unlocking the true power of robotics. We need a paradigm shift that includes hardware too.
This is the thinking behind ARIA’s Robot Dexterity programme, which focuses on robots’ bodies. What if, instead of just making robots smarter, we focused on making them more nimble too? Following this approach means redesigning robot bodies: embedding them with sensors, creating new materials, inventing new actuators, and integrating these closely with the software that controls them.
Ultimately, regardless of the technological path taken, it’s the robot’s capabilities that matter. If we focus on rewarding outcomes over specific methods, we open the door to unexpected advancements that deliver tangible results.
Challenge prizes are at their best when they set clear outcomes but are agnostic on methods – so it should come as no surprise that we think there’s potential for an ARIA robotics prize to set an ambitious goal and catalyse further progress.
A future ARIA robotics challenge prize
Today we’re publishing our prize prototype – our initial draft for how we think the prize could be structured and focused. You can read it here:
We want to develop a prize that will change the conversation globally about robotic capabilities. We want feedback to make sure we scope something that truly pushes the boundaries of what is possible, brings in radical thinkers and inspires ambitious work.
Over the coming weeks we will be engaging with experts and stakeholders to refine our thinking. We may post an updated draft if our thinking changes substantially in light of feedback.