Humanoid robotics is a field where people could encounter robots in real life. Created to help, assist, and work, humanoid robots have entered the domain of AI in the past years. Still, there are some considerable challenges left to improve on hardware and software to become a viable product.
Creating humanoid robots – robots that act like human beings – is much about acceptance. Designed to interact with people, humanoid robots bring along some significant advances in robotics. Up until now, there is a problem that is hard to solve, though. To be genuinely humanoid per definition, these robots need to be able to move forward up on legs, and can't use much-advanced locomotion techniques like wheels or quadrupeds. While there are companies that mainly focus on the upper body part, psychologists argue that this was violating the acceptance necessary to create an in-depth relationship of some kind. So, this difference in locomotion seems to be relevant and could become the bottleneck for improving on the credibility of humanoid robots. The whole concept of humanoid robotics bases on the appearance of human beings. Why is that?
"people would interact with robots more naturally when these robots resemble human beings"
Mainly because humanoid robots are designed to be integrated into our daily lives, workplaces, homes, and families, to help, work, and assist everywhere we go. Psychological studies suggest that people would interact with robots more naturally when these robots resemble human beings. Moreover, a more human-like appearance raises the expectations of robots to be more intelligent. That's why service robots are designed to look more human than industrial robots, although they do not necessarily need to look exactly like humans.
What seems contradictory refers to a widely known theory called "uncanny valley". It says that people sense positive feelings towards robots if their physics are familiar to humans, but, at the same time, people disapprove of robotic life forms that look too similar. Latest assumptions see a different dimension behind this theory, though. It might not be about looking too similar, but not feeling similar enough in nuances. As a result, robot creators mostly keep the appearance – especially the face – a safe distance away from human beings. Critics complain that there is too much research spending on looking and moving like human beings when the goal should be utility. However, to truly fulfill the purpose of serving as part of our daily lives, this acceptance appears to be crucial, and these research dollars seem to be inevitable.
While robotics has already become an integral part of industry 4.0, there is no business reality for humanoid robots yet. That's not because of a lack of market applications, though – there is a wide range of future applications in daily lives:
Most of the R&D dollars run into healthcare and military purposes. These sectors are said to be potential drivers for this challenging market and could increase sales by 30-45% until 2025. Demographics will lead to a future where assistance by robots would be of great help. Also, governmental programs see something in humanoid robotics and enable research on that matter, e.g., by US agency DARPA (Defense Advanced Research Projects Agency). Nevertheless, these investments have to materialize someday. That's why the industry is under pressure to deliver on people's expectations. Take frontrunner Boston Dynamics, which has faced the common struggle for profit and was then sold by Alphabet (Google) for reportedly 100 million dollars – not confirmed though – to SoftBank Group, a Japanese tech conglomerate that is also producing robots.
"Working on humanoid robotics is still way too expensive due to uncertain market perspectives, low numbers of pieces, complex processing, and high material costs."
Working on humanoid robotics is still way too expensive due to uncertain market perspectives, low numbers of pieces, complex processing, and high material costs. Robots run several motors, sensors, actuators, power packs, computer parts and software systems – so, a humanoid robot can cost up to seven figures in dollars. Boston Dynamics' world-renown Atlas, for example, is said to cost around 2 million dollars. "Entry-level" robots go from about 70,000 dollars for RoboThespian (Engineered Arts) and a reported sum of 200,000 plus dollars for iCub (Italian Institute of Technology).
Experts say, there is another factor that could slow down growth in humanoid robotics. There is still enough reservation on people's minds about robots taking over society and stealing jobs. So far, however, these robots are designed to carry out tasks that people don't want to do or can't do. Plus, technological progress will likely create new jobs – and if it's for handling future robotic mechanisms.
As mentioned before, one of the major challenges of humanoid robotics is with setting up bipedal locomotion. A robot to navigate around homes and unstructured environments autonomously accounts for incredibly difficult mechanical engineering. Pathfinding, coping with dynamic situations, and recovering from falls require sophisticated mechanisms like self-balancing capabilities and gravity adjustments. There is a list of companies that have come far in this respect. To name a few:
"people seem to be scared of human-looking robots using AI in our environment"
There are a few more robotics companies e.g., from Japan, Korea, and Europe, that created considerable progress in the fields of dexterity and versatility. And, there are plenty of challenges left to solve. Besides the bipedal issue, researchers have to develop ways how to power the robots in a not tethered way, how to cool down the heat produced by the robots' mechanics, and how to install a stable AI skill set. Especially concerning the last aspect, people seem to be scared of human-looking robots using AI in our environment. So, what future of human-robot interaction do we face?
The interactions between humans and robots do, so far, not necessarily mean fully AI-based encounters. There is a wide range from remotely controlled robot collaboration up until human-robot communication by using full-scale AI.
One of the most famous robots for human-robot interaction is Pepper, created by French company Aldebaran Robotics, later purchased by SoftBank Group. It is a semi-humanoid robot with wheels instead of legs, that can read emotions and communicate on a basic level of intelligence. Pepper was sold more than 10,000 times around the world and can be met everywhere on trade shows, on television, and at marketing gigs.
Human-robot interaction is not limited to customer experience, though. One major field will be collaboration in industry sectors. Take another semi-humanoid robot on wheels: the Armar-6 from the German Karlsruhe Institute of Technology. This cobot – collaboration robot – mainly focuses on assisting in specific tasks, and on learning and understanding where assistance is needed. Some people might argue why not make a cobot with more arms and eyes on the backhead for truly fulfilling utility purposes. However, even though this makes sense to some degree, experts still weigh the interaction factor between humans and robots more than utility aspects and estimate the humanoid resemblance to be crucial for natural human-robot interactions and collaborations.
The most sophisticated humanoid robot in terms of AI-based communication might be Sophia from Hanson Robotics. Sophia has become a technology pop-star around the world so that we can focus on the AI aspect. Some experts, many of them being competitors, state that Sophia is not entirely based on AI and operates more like a puppet reacting to keywords. What might be interesting with Sophia, though, is the opening to create AGI, Artificial General Intelligence. Sophia runs on three different control systems: Timeline Editor, Chat System, and AI platform named OpenCog. While the first two software components allow for stable communications, OpenCog allows for learning from answers and experiences.
"this robot has opened up many minds for various other topics, like AGI swarm intelligence and ethical matters"
Interestingly, Sophia-like robots could get connected via wifi and update their learnings in real-time – creating an intelligence level way beyond human competence. That's a glimpse of what we can expect of the future of AI-based systems and how this might become way more than a business game-changer. So, even if Sophia was currently only a more sophisticated puppet, this robot has opened up many minds for various other topics, like AGI swarm intelligence and ethical matters.
Ethical questions didn't only arise because of Sophia having received citizenship from Saudi-Arabia and Sophia stating that humanoid robots should be able to have families too. Most of the ethical questions derive from that well-known apocalyptic thought of AI-based humanoid robots taking over control of human beings and thus becoming extremely dangerous. Prominent representatives like Stephen Hawking or Elon Musk are warning from irresponsible use of AI, so the three laws of robotics, once science fiction, could become an essential prerequisite of creating. Considering the fact that AI development seems to be unstoppable, it should be of importance to educate on AI, instead of just warning from that matter. And, to be realistic, AI and humanoid robotics technology are not quite there yet. What is officially known, there is still much work to do.
So, do we see a Robo age incoming any time soon? At least everybody senses the advantages that might come along with robot-generated leverage. However, while we acknowledge automation and AI entering industry-scale applications, the humanoid robot business seems to be stuck in a stadium where robots are considered to be fun, but not truly helpful so far. Companies are vying for opportunities to put their products into use. However, to come forth, further developments will eat loads of additional research dollars.
"Humanoid robotics is still considered a very interesting high-risk-high-outcome market."
As for the market perspective, the Asian market has got the potential to show the most significant growth factor, primarily due to Chinese and Japanese robotics companies. Maybe the US market will be able to catch up a bit. Other regions of the world might need to boost their efforts vastly to become competitive. Humanoid robotics is still considered a very interesting high-risk-high-outcome market.
You cannot say whether humanoid robotics is the next logical step on the technology ladder or just wishful thinking made up of what people would like to foresee. Chances are, the latter assumption leads to the first: humanoid robotics could at least become a self-fulfilling prophecy. So, in the end, humanoid robotics might become a business reality sooner or later. I bet it'll be sooner.