Small, lightweight and designed to be inherently safe for close human operation, cobots are an ideal entry point for many companies looking to embrace robotics.
Conventional industrial robots are impressive machines, carrying substantial loads at lightening speed with great precision.
But they are expensive, inflexible, complex to programme – and liable to take your head off if you stand too close.
“What might start out as a £20,000 or £30,000 industrial robot arm can cost well over £100,000 to finally implement on the factory floor,” says Peter Machin, field application engineer at Rethink Robotics. “Then it takes up enough space to build a house on.”
The cobot advantage
Collaborative robots – cobots – are different. Lightweight and compact, they can work right alongside humans thanks to soft skin and force-limiting sensors that keep the energy of any collision at a safe level.
With no need for protective guards and simple to programme, deployments cost far less.
The big idea behind cobots is simple enough – efficient working. Man and machine complement each other, with the human using its big brain for complex tasks while the robot rapidly processes the product or prepares it for human attention.
“What we want to automate now are the repetitive human tasks,” says Machin. “They tend to be straightforward: pick that part up and place it somewhere else.”
Cobots can handle lifting and take over monotonous or dangerous tasks, leaving the more dexterous – and rewarding – jobs to the humans and helping them to avoid injury into the bargain.
One MIT study found that cooperation reduced human idle time by 85%, making cobot-human working more productive than purely human or robot teams.
“The robot is really like another fellow worker,” relates Machin. “It’s not seen as a threat and we’re seeing it actually create employment.”
Flexibility is another cobot advantage. Instead of being fixed in place forever, you can wheel cobots to the end of another line and, with pre-programming, get them working again in minutes.
Machine-tending and parts assembly are proving popular applications across sectors like automotive, plastics and electronics.
Real Digital, a London-based print and packaging company, allocated a particularly dull job to its first Universal Robots device: picking and stacking cardboard boxes. It cost £22,000, about the same as a year’s wages for a human employee.
At Jaguar Land Rover’s Solihull plant, a KUKA LBR iiwa and a human operator work together. While the cobot applies sealant to a new part, the operator picks up an already sealed part, fits it to the vehicle and prepares the next component for the cobot’s attention.
“A standard robot could handle this application but there’s not enough room on that part of the production line,” says Adam Webb, applications engineer at KUKA Robotics. “The cobot can work in a more confined environment.”
Like many other clever ideas, cobots are far from new. They first appeared more than 50 years ago, following research funded by robotics pioneer General Motors.
GM’s first Intelligent ‘Assist Device’ was a movement-limited hoist that helped Detroit car plant operators shift heavy loads easily.
Today’s typical cobot is much smaller, comprising one or perhaps two aluminium or magnesium arms featuring the force-limiting sensors described above.
Back in 2009, Universal Robots demonstrated this format’s commercial potential with the 5kg-carrying UR5. It now offers two other six-axis arms with different payloads: the UR3 and UR10.
Rethink Robotics was another innovator, releasing its Baxter model in 2012 and following up in 2015 with the faster, stronger, and more precise Sawyer. It boasts a vision system that can find parts and a seven-axis arm with force sensing that lets it “feel” its way into features.
Though many other manufacturers like KUKA, ABB and Fanuc have released cobots over the past five years, Universal Robots is the runaway market leader with around 17,000 installations.
As production volumes grow, Barclays Equity Research predicts cobot prices will decline by 3 – 5% annually through 2025.
At £4,990, UK start-up Automata’s Eva will be the lowest priced cobot yet when it ships later this year. It even has an option to pay a monthly subscription rather than a one-off fee.
That compares to a price of around £25,000 for a UR5. Adding a simple two-finger adaptive gripper plus installation costs would see that UR5 running for about £38,000.
But when the usual rule of thumb with industrial robot deployments is to triple the buying price, it’s little wonder senior management are eager to try cobots out.
Cobot learning curve
GKN is one example of a company investing in cobots, with its CEO allocating £2m in 2016 to fund 14 trials. Most of the projects delivered productivity increases and paid back within a year, though there were some setbacks.
“The applications ranged from simple pick and place to very sophisticated assembly,” says Mohammed Zameer, vice president of Global Manufacturing at GKN Driveline.
“But cobots are not just plug and play, no matter what some suppliers might have you think. Don’t underestimate the development time for complex applications. We went through a very long learning curve on some of them.”
To avoid the need for a camera system, parts need to be presented to the cobot in a uniform manner in a known position, ideally in a tray. Zameer recommends designing the new process and then simulating it to speed development.
“The biggest limitation was speed,” he says. “We found that if we tried to maintain normal speeds, we experienced a lot of slips and drops.”
Safeguarding the operator against those and other hazards needs to be considered from the start. New guidance on cobot installations arrived in 2016 (ISO/TS 15066), laying out how to design and risk assess both the cobot and the workspace shared with humans.
Correctly restricting movement, speed and payload are critical parts of the process.
“There are two main hazards, a collision in free space and crushing,” says KUKA’s Adam Webb. “To minimise crushing and trapping hazards, you design the cell so the cobot keeps away from the operator.”
If the cobot is carrying a hot part, that needs to be shielded. If it’s holding a sharp-edged PCB, then that should be turned during movement to avoid it acting like a knife.
Collaborative processes need collaborative grippers too. “Unless you have a sensor inside the gripper, it will not know whether it is gripping a metal part – or your hand,” says Zameer.
National HSE legislation has struggled to keep up with cobot progress. GKN found that standards varied widely, with some countries still requiring guards for all robots. In others, local officials set the allowable extent of collaborative working.
In the UK, the ISO robotics and cobotics standards are applied and these are widely seen to be reasonable, if conservative.
The HSE has so far produced one inconclusive report (RR906) in 2012. Its Centre for Shared Research has a current project looking at best practice and future improvements for cobot safety.
“The safety side is still playing catch up with the technology,” says Mark Gray, UK sales manager at Universal Robots. “The ISO standards are written for Stephenson’s Rocket when you’re actually working with a Pendolino.”
Teach your cobot
Robot programming is complicated and the UK only has a handful of extremely well-paid programmers. Cobot training, as it’s commonly known, is so simple a child could manage it.
Cobots’ software is their most innovative attribute by far, letting users set up processes in hours instead of days or weeks.
Some cobots also feature ‘lead-through programming’. To record the path that the cobot will follow, operators literally take them by the arm and show them where to go. But this simplicity can have its drawbacks.
“Because they are so easy to programme, every operator wanted to make a change as soon as they found a problem,” notes Zameer. “In doing that, they can create a hazard. That was something we learnt the hard way.”
Customising end effectors and their control code to suit individual processes is traditionally one of the biggest robotics challenges. Though cobot arm tooling may still be costly, innovative software can make it much easier to plug in standard tools.
Universal Robots’ UR+ scheme is a good example, with 12 manufacturers currently standardising their connectors and code to gain UR+ accreditation. “We certify that it will work with the robot and integrate their software with our software,” explains Gray.
Sophisticated force-sensing, seven-axis arms, integration with automatic guided vehicles: cobot development has moved fast. But basic human tasks like looking through and selecting correctly-sized fasteners from a box full of nuts and bolts still flummoxes them.
One study found that it took a robot 20 minutes to fold a towel. More and better sensors along with some form of machine learning are needed to attain human-like hand-eye coordination, extreme force sensitivity and dexterity.
“Learning and adapting how it picks things up in the way that a human does, that would really change the game,” says Machin.
Improving human-cobot interaction is another research focus. To help workers predict the motions of its Sawyer cobot, Rethink Robotics made its eyes (displayed on its head-like screen) move in the direction of the upcoming hand movement.
Sensors on workers’ overalls can help the cobot detect nearby humans more precisely.
Recent MIT research has also employed wearable sensors, using human-generated brain signals to instantly alert cobots sorting objects to an error.
However, today’s technology is proving more than sufficient to attract users. Energias reports that the global market is growing at over 50% annually, rising from $177.2m in 2016 to well over $4bn by 2023.
The new normal
In the UK, cobots are now doing everything from stacking boxes, painting and welding to gluing on car door liners and basic quality-control inspections. GKN will kick off another 11 cobot projects in 2018.
Cobots have an obvious role in helping to fill the UK’s skills gap, freeing up human capital for more productive roles. They can also help UK manufacturers compete against rivals from low-cost markets and cope with the Brexit labour shortage – long predicted and now real.
“In the past month, I’ve sold six cobots to companies that are struggling to find staff because Eastern Europeans either aren’t coming over in sufficient numbers or are returning home,” says Universal’s Gray.
Cobots’ lower prices, simple programming and versatility let more companies take part in the robotics revolution. On the UK’s production lines, working alongside a cobot colleague looks like becoming the new normal.
Reporting by James Lawson.