There is considerable misunderstanding around the role of AI in robotics. Human-like robots with super intelligence are standard fare in entertainment, while terms such as ‘robotic process automation’ and ‘bots’ refer to software applications with no physical robot.
To separate fact from fiction, the IFR has just published ‘Artificial Intelligence in Robotics’ , in consultation with members and experts. The paper looks at current applications of AI in robotic applications, future directions, and safety and certification considerations.
Artificial intelligence opens up new possibilities for robotic automation. The growth of customised production, with frequently changing products, orders and stock, and the rapid rise of e-commerce, have made variability and unpredictability a common feature of manufacturing and logistics sectors. Uncertainty is also inherent in public environments where healthcare or assistance robots, for example, are at work.
In these situations, AI enables robots to sense and respond autonomously to their external environment in real-time, for example identifying objects to be picked from an unsorted bin, or automatically identifying welding points on a new part.
AI also holds potential for reducing the time and resources needed to programme and re-task robots. Programming and integration account for 50-70% of the cost of a robot application. This, together with the time and cost of re-tasking a robot to the requirements of a new production run, have made automation economically unviable for many small-to-medium-sized companies (SMEs), or for larger manufacturers and wholesalers with high product variance. Experts estimate AI could ultimately halve the time required to programme a robot, as well as significantly cut the time needed to re-task it.
It’s important to note, however, that AI is not necessarily a prerequisite to enable robots to respond in real time to their environment. Many ‘pick-and-place’ applications, in which the robot identifies an object to be picked and determines how to approach and grasp the object, do not require AI. However, the greater the level of variability and uncertainty, the more likely it is that AI algorithms will bring cost-benefits over traditional, deterministic programming. For example, while mobility does not require AI, capabilities such as determining whether a person encountered by a mobile assistance robot is an adult, older person, or child, and then responding accordingly, would either be impossible, or require unrealistic amounts of computing power to achieve, without AI.
The fact that AI algorithms enable robots to act autonomously raises questions around safety. However, robot applications comprise many different levels of software. Currently, the safety layer is hard-coded – for example, ‘stop if an object is less than 10cm away’ – and does not use AI. This may change in the future – for example AI may be used to determine how fast a robot should slow down, based on the trajectory of the object or person in front of it.
European authorities are currently reviewing regulation on AI in robotics. The global robotics industry is concerned about the current drafts of the European Machinery Product Regulation and the AI Act, particularly with regard to the proposed requirement for mandatory third-party certification of AI-enabled robots (vs. manufacturers’ self-declaration of conformity with safety standards as is currently the case). This would impact any company selling robots on the European market. In the long run, the new regulation will disadvantage European companies, especially SMEs and start-ups. The IFR is also worried that the relevance of international standardization - a cornerstone for the safety of robots around the globe - might shrink if the European Commission mandates the development and adoption of EU-specific technical specifications. The IFR therefore calls on European policymakers to amend both drafts to balance the protection of citizens with the market’s need to adopt new technologies and ensure a level-playing field for companies.