Research Publication – Artificial Intelligence – Robotics – The Long Dawn

Robots have great promise, but it is going to be a very long wait. As usual with systems powered by AI, the problem remains causality, and there is still no good solution. Robotics have to be able to navigate the human world, and be capable of multiple tasks, making this a hard problem to solve. RFM thinks that free-ranging robots will appear first in factories and that autonomous driving is a lead indicator of technical progress. RFM continues to think that neuro-symbolic AI offers the best opportunity to solve or mitigate the causality problem.

• The Long Dawn: Recent advances in AI have brought robots onto the horizon of public consciousness, and interest in the sector is rising. Marketing videos show robots doing backflips and other tricks, but these are not representative of any market reality as there are still many technical problems to be solved.
• Great Potential: Robotics has very large market potential and RFM thinks that the market for domestic house robots alone could be bigger than the entire automotive industry in time.
• Causality: once again lies at the heart of the problems, as statistical-based systems are being used to teach robots to move around using legs and to perceive their environment. This means that robots today encounter great difficulty as soon as they encounter something new.
• General Robotics: is the holy grail. This is where a single model can be used in any robot without needing extensive retraining. This is a requirement for economic viability. Currently, every time a change is made, retraining is required, making robots very difficult and expensive to make and maintain. RFM sees this as a major hurdle to mass market adoption.
• Environment: is also problematic, as deep learning is unable to deal with the virtually infinite variability of people’s homes and the human world as a whole. This is similar to the problem encountered in autonomous driving but in this case, the problem is more severe.
• Factories First: RFM thinks that free-moving robots are likely to be deployed in factories first, as their environment and tasks will be far more limited than the open environment of a house or outside in the street. With a high ROI, companies will also be in a position to pay the higher prices that will be demanded in the early years.  
• Autonomous Vehicles: are comparatively simple robots. Hence, the problem of autonomous driving is likely to be easier to solve than general robotics, meaning that it will come to market first. Hence, it serves as a good indicator for robotics in this early stage of the market.
• Neuro-symbolic AI: is still the best option to mitigate the shortcomings of statistics-based AI systems in RFM’s opinion.