Our cities, streets, homes, and businesses are built for beings that walk on two legs (biped). From stairs to the shape of hallways to the placement of kitchen cabinets, all have been designed for bipeds. The fact that a society of two-legged creatures designed everything around them for bipeds is so obvious most people don’t even think about it, but it becomes a serious issue when talking about the future of robotics.
If you want robots to be able to fully operate within our society, replace certain human jobs, and be able to navigate all parts of our buildings with the same ease humans do, perfecting robotic walking may be the best way to do that. The problem is that building a robot with wheels is fairly simple, but building a machine to walk on two legs is a complex technological challenge.
This article examines the cutting-edge developments of four of the top bipedal robotics companies, including:
- Boston Dynamics
- PAL Robotics
- Agility Robotics
For each company, we’ll cover current bipedal developments, predictions and projections from executives (i.e. “when will robots walk?”), and challenges for the bipedal developments ahead.
Most people likely take walking for granted, but it is in fact a highly complex process. A single step requires you to use nerves signals from your entire body to figure out the position of the foot, the slope of the ground, and the relative location of the rest of the body.
Your brain then performs numerous calculations in under a second to determine how to maintain balance and to have dozens of muscles make tiny corrections to keep you upright. In the video below, University of Michigan’s Dr. Jessy Grizzle explains some of the principles of bipedal movement in it’s MABEL robot:
Jonathan Hurst’s 2016 TEDxOregonStateU talk explores some of the challenging dynamics of replicating walking and running in robots:
People are often unaware of all the unconscious effort that goes into simply walking, but robotic engineers are painfully aware of the complexity they are trying to replicate.
(Readers with a more broad interest in robots may want to read our general guide article called on machine learning in robotics.)
Top Organizations Working on Biped Robots
Two countries that have been driving forces behind the development of two-legged robots are Japan and the United States. The shrinking population of Japan has in part caused the nation to focus on innovation in robotics as a way to remain highly productive with fewer people. In the United States the military’s strong interest in robotics has helped drive innovation and early investment. The US’s Defense Advanced Research Projects Agency (DARPA) has been aggressively funding and encouraging robotic research for years.
Below, we’ll assess some of the US and Japanese companies leading the field:
Honda has long been a leader in humanoid robotics. Honda has been working on the development of robotic bipedal locomotion since the 1980’s. The first big demonstration of their technology was the first ASIMO (stands for “advance step in innovation mobility”) in 2000. The humanoid robot was capable of slowly walking without a support system.
Honda has been steadily improving their ASIMO line over the years and in 2011 created their first “autonomous” ASIMO. It added more advanced balancing capacity, external recognition capability and the ability to autonomously determine its next behavior without direct input from the operator.
The most recent version of ASIMO is now capable of going up and down stairs, walking forward and backward, dancing, hopping on one foot, and running 5 km per hour. Honda has been steadily improving the system’s performance over the years, making it faster, more stable, and more agile. ASIMO is just over 4 feet tall, 119 pounds and can operate for roughly an hour.
ASIMO is designed to be a helper for people in need — as a potential nurse, office assistant, or butler. At the moment, though, ASIMO’s only jobs have been effectively public relations related. The robot has mainly been used to promote Honda and in limited high profile roles like being a greeter at a museum. Back in 2014 Honda’s chief engineer said it would be another 10-20 years before they mass produce robots for households.
Boston Dynamics (SoftBank Group)
Boston Dynamics developed as a spin-off from the Massachusetts Institute of Technology and received funding from Defense Advanced Research Projects Agency (DARPA) to develop BigDog, a quadruped robot. The company was bought by Google X in 2013 and sold to Japan’s SoftBank Group earlier this year. As part of the same deal, Google also sold SoftBank SCHAFT, another company working on biped robots.
Boston Dynamics’ most recent version of their Atlas robot was unveiled in 2016. The robot is capable of working on uneven ground, lifting boxes off the ground, and getting itself back up if it’s knocked down. The ability of Atlas to be able to work around outside in a real world forest over unpredictable ground is a big step forward for the technology. Atlas stands at 5 feet tall, weighs 165 pounds and can carry a payload of 25 pounds.
Softbank is already heavily invested in bringing human-like robots to market. In 2015 it started selling Pepper, a wheeled humanoid robot capable of recognizing the principal human emotions, which is currently used in over a hundred stores as part of their customer service. The use of retail robots is currently relatively limited, but pursuable adding walking capacity to robots like Pepper would expand the areas and ways they could be used.
The Spanish company PAL Robotics is a leader in humanoid robots and has been building biped robots for over a decade.
They recently announced they would be selling the TALOS – a biped robot capable of walking up to 3 km/h and able to move over uneven ground. The 1.75 meter tall robot weighs 95 kg and can lift up to 6 kg in each arm.
The robot will start out as a tool for research ways to improve robotics, but the hope is that within the next five years the robot will be able to actually work side by side with people in industrial applications such as factories. The longer the goal is to develop Talos to do work that could be too dangerous for humans, like going in to do repairs after an incident such as the Fukushima nuclear disaster.
Agility Robotics is a new startup and a recent spin-off from Oregon State University. The projects launched was aided by a $1 million grant from DARPA.
Their biped robot Cassie instead of trying to copy how humans works, mimics the way many birds walk. The robot is even named after the large Cassowary bird, which it vaguely resembles. Cassie is capable of dynamic walking, running, sitting, and stabilizing itself when pushed.
The new company is already selling Cassie to other academic and research institutions. Its long term vision is to have robots capable of assisting in disaster areas and armies of robots that bring packages right to your door from autonomous delivery vans. The vision is self-driving vans will do the driving, and their robot legs will bring the packages down the driveway or up the stairs to your door. As co-founder Jonathan Hurst has pointed out, “flying drones are not going to be delivering 50 pound bags of dog food.”
Do We Even Need Biped Robots?
One practical consideration that could really hold back the development and commercial deployment of biped robots is the fact that they might be unnecessary.
Wheeled robots do have a serious drawback when using them in many real world applications. Wheels inherently have a difficult time going up stairs, clearing curbs, or going over fallen trees or other objects. As a result, there are many possible situations where a robot with legs would be superior to a robot with wheels, but that doesn’t necessarily mean two legs are better than four. Dogs are more than capable of navigating our cities and buildings. In addition, our recent article on AI and robotics in food services didn’t list a single example of biped robots.
Four legs are inherently more stable than two, which makes four legged robots easier to build. For example, Boston Dynamics quadruped robots like Spot and SpotMini can move over uneven ground, go up stairs, and run at a fast pace. They are clearly more nimble than Boston Dynamics’ biped Atlas robot at the moment. While the same weight, Spot can also carry a significantly heavier payload than Atlas. We asked TechEmergence CEO Daniel Faggella to comment:
“Biped robots have made amazing progress in the last 10 years, but it seems unlikely that they’ll be the most popular or useful robots in commercial or industrial applications in the short term. Robot dexterity and balance are extremely challenging problems, and until they’re solved, wheeled or four-legged robots seem to have the advantage for usefulness in the home or on a job site. I don’t suspect biped robots to be common in either domain within the next half decade.” – Daniel Faggella, CEO at TechEmergence
As their video shows, SpotMini is capable of navigating around a home, and a robot arm theoretically allows a four legged robot to perform the same tasks you would want a humanoid robot for.
Humans aren’t bipedals for any reasons related to efficiency or simplicity. Walking on two legs was an evolutionary trade-off made to free up hands to grab things and manipulate tools. When building robots from scratch, though, there is no need to follow the same biological constraints our evolution did.
The future of delivery robots, home aid robots, and rescue robots might not be robots that look like humans but ones that look more like centaurs. Picture a four-legged robot with arms.
There are still some instances where biped robots might make more practical system or be preferred. Biped robot could potentially move easily navigate very narrow spaces or go up ladders. Possibly the biggest reason for biped robots is customers might find a more human like machine more comfortable to interact with. That is a less important concern with something like a delivery robot but more important for a robot that might be used in a nursing home or retail store.
Concluding Thoughts on Biped Robot Timelines
Building a biped robot capable of human-like movement in real world settings is a surprisingly complex task. It is a challenge that companies and engineers have been working on for decades.
There are clear practical benefits of robots capable of moving around on legs, be it two legs or four legs. If the technology is perfect, it could be truly transformative, changing everything from delivery to combat, to health care services and how we live our lives. So we can expect companies and militaries to continue spending money on researching the challenge, but the technology remains squarely in the research phase for the time being
While significant progress has been made and there are currently some impressive robots, none of the top developers in the field have built a robot that can match a human’s ability to move around on two legs. As a result, at this moment, no companies are currently selling biped robots for wide commercial use or are expecting to in the immediate future.
The first people really benefiting from the research on biped robots have been the disabled. The research and technology Honda developed for ASIMO lead it to create the Honda Walking Assist technology.
While we haven’t been able to build robots that walk as well as people, we have been able to make individuals walk better.
Related Robotics / Vehicle Timeline Articles
Our editorial mission at TechEmergence is inform business leaders about the AI innovations that will impact their industry, and much of our robotics and vehicle innovation coverage has involved timelines and predictions. If transportation is an important part of your business – you may benefit from our other recent timeline pieces.
- Autonomous Car Timeline Predictions (GM, Ford, Honda, and more)
- Self-Driving Truck Timeline (Daimler, Volvo, and more)
- Autonomous Boat Timeline (Rolls-Royce, Kongsberg, Yara and more)
Header image credit: Auto Express YouTube Channel