Research inspiration sometimes comes in unexpected guises. Who would have thought, for example, that academics researching the mechanical act of robots ‘grasping’ objects would find the keys to begin unlocking the mysteries of robotic movement in existing research stretching back to the 1950s in a bone and joint journal?
How could insights about how humans manipulate their surroundings guide Dr. Valerio Ortenzi and his team of international collaborators in devising a paradigm for robotic grasping that could be a ‘game changer’? How might an insight into bone and joint structure propel forward our understanding of robotic manipulation – potentially speeding our progress to a time when we have robot assistants in our homes and workplaces?
Get updates from Quest directly to your inbox.
Subscribe to Quest for our research stories.
Grasping – the first step to success
Humans do not grasp objects for the sake of it. As toddlers, we play around with objects to understand the world around us and how our bodies function. However, as adults, grasping has a clear purpose - the significant first step to a successful manipulation.
Grasping is an action perfected long ago in nature but one which represents the cutting-edge of robotics research. Defining a successful grasp is a tricky problem; humans continuously adapt while performing grasping, whilst robots do not yet possess this reactiveness and a grasp is classified as successful if the object is held stably in the robotic equipment for an arbitrary time.
Dr Ortenzi argues that robots need to know the reason why they are doing a job if they are to safely work alongside people in the near future. In simple terms, this means machines need to understand motive the way humans do, and not just perform tasks blindly, without context.
Change approaching for robotics’ world
Based in the National Centre for Nuclear Robotics at the University of Birmingham, Dr Ortenzi has worked with researchers from the Centre of Excellence for Robotic Vision at Queensland University of Technology, Australia; Scuola Superiore Sant’Anna, Italy; the German Aerospace Center (DLR), Germany; and the University of Pisa, Italy to deliver research that could herald a profound, but necessary, change for the world of robotics.
In a recent paper published in Nature Machine Intelligence, they argue that the shift in thinking will be necessary as economies embrace automation, connectivity and digitisation together with increased levels of human–robot interaction. Robotics researchers and engineers will have to apply ‘purposive’ principles to robot grasping if we are going to achieve the necessary pace of technological progress.
Image credit: Elastico Disegno snc and Prensilia Srl.
‘Dumb’ machines need learning
“Most factory-based machines are ‘dumb’, blindly picking up familiar objects that appear in pre-determined places at the right moment,” explains Dr Ortenzi. “Getting a machine to pick up unfamiliar objects, randomly presented, requires seamless interaction of multiple, complex technologies.
“Vision systems allow the machine to see the target and determine its properties; advanced AI calculates billions of possible positions of a multi-joined arm and plans a path without collisions; sensors in the gripper prevent the robot inadvertently crushing an object.
“However, even when all this is accomplished, what has traditionally counted as a ‘successful’ robotic grasp may be a real-world failure, because the machine does not take into account what the goal is and why it is picking an object up.
“Imagine a factory robot picking up an object for delivery to a customer and successfully executing the task, holding the package securely without causing damage. Unfortunately, the robot’s gripper obscures a crucial barcode, which means the object can’t be tracked and the whole delivery system breaks down because the box has been grasped the wrong way.”
Grasping with purpose leads to success
By the 1950s, studies in human physiology had already established grasping as a purposive task with the intended activity influencing the pattern of grip and choice of hand shaping dependent on the manipulation task to be performed. Grasps fall into three main classes: precision grasps, such as grasping a little ball; power grasps, like grasping a hammer; and those combining both these - intermediate grasps used, for example, to grasp a pen when writing.
There are also five factors determining the choice of grasp:
- Object features - shape, size and embedded function;
- Force and mobility – depending how the object is used once grasped;
- Size and power constraints of the human hand;
- Prior experience - habits of the person grasping; and
- Chance - affecting approach to the object and dynamic grasp.
Two important additional factors are safety and social convention. Dr Ortenzi explains further: “Imagine asking a robot to pass you a screwdriver in a workshop. Based on current conventions, the best way to pick up the tool is by the handle, but that could mean a hugely powerful machine then thrusts a potentially lethal blade towards you, at speed.
“Instead, the robot needs to know the end goal – in this instance, to pass the screwdriver safely to its human colleague, in order to rethink its actions. What is obvious to humans must be programmed into a machine. This requires a profoundly different approach. The traditional metrics used by researchers, over the past twenty years, to assess robotic manipulation, are not sufficient anymore. In the most practical sense, robots need a new philosophy to get a grip.”
Header image credit: Elastico Disegno snc, Tommaso Pardi and Prensilia Srl.
Discover more stories about our work and insights from our leading researchers.