Robots play an important part in our everyday lives. Non-autonomous systems can be found in industry, surgical theatres, and even our homes, and more autonomous robots are integral to space and deep-sea exploration.
Dr Sirko Straube and Professor Dr Frank Kirchner from the German Research Center for Artificial Intelligence (DFKI) are seeking to translate this greater autonomy of space and maritime applications to the human sphere.
Read the original research: doi.org/10.1007/978-3-031-11447-2_4
Read more in Research Outreach: doi.org/10.32907/RO-140-6297744387
Image Credit: Adobe Stock / Copyright DFKI GmbH
Transcript:
Hello and welcome to Research Pod! Thank you for listening and joining us today.
In this episode, we look at the work of Dr Sirko Straube and Professor Dr Frank Kirchner from the Robotics Innovation Center of the German Research Centre for Artificial Intelligence, or DFKI, who seek to translate the greater autonomy of maritime and space applications to the human sphere, opening possibilities for fruitful interactions between man and machine.
The term ‘robot’ often evokes images of the highly intelligent and humanoid mechanical systems described in books, films, and popular media. The reality of modern robots is more complex, with robotic systems such as food preparation devices, smart vacuum cleaners, rehabilitation exoskeletons, drones, and industrial robotic arms already in use, and more applications proposed in almost every sphere of human experience.
Computer algorithms designed to perform tasks normally requiring human intelligence, like speech recognition, decision-making, and language translation, are collectively referred to as artificial intelligence or AI. AI has become increasingly advanced in recent years, with algorithms even able to beat human players at strategic games, such as Go and chess, through trial and error learning. The most recent advances also include the likes of AI systems, such as ChatGPT, that boost the application of AI in everyday life and also function as a booster for robotics technology.
Robots are made of real-world hardware – a body, mechanical components, electrical circuits, sensors to monitor conditions and flag when changes occur, and actuators to receive signals and perform actions. All of these are connected through a composition of software programmes. Modern robots are often developed in tandem with integrated AI software to produce intelligent robotic systems with varying degrees of autonomy.
In collaboration with colleagues from social sciences, Dr Sirko Straube and Professor Dr Frank Kirchner from the DFKI used the Delphi method of progressive and iterative questionnaires to determine how people perceive different types of robotics applications, from household tasks to space exploration. Participants’ responses to robotics and AI were overwhelmingly positive, with only one in five viewing them negatively, and only one in ten seeing robots as unnecessary for society.
However, a few considered AI and robotics to be reliable and error-free. Responders were also much more critical of robot use in private homes and care than in the context of deep-sea or space exploration. The team suggests this might show a preference for using robots far away from human settings. Emotional considerations could also be a factor, as care-giving is often associated with empathy, a characteristic not commonly associated with machines.
Autonomy is a key aspect of any robotic system. Developers tend to classify robots as non-autonomous if they are fully controlled by human operators, semi-autonomous if they use AI to operate independently in certain tasks but require human help in some cases, and fully autonomous if they can plan, replan, and react appropriately to their environment in real time, without the need for human input. In reality, most current systems still involve a degree of human support. The level of autonomy depends on the environment, the task in hand, and the context of the operation.
Straube and Kirchner stress that autonomy is useful in highly uncertain scenarios. When a robot is placed on the Moon or Mars, it will naturally be exposed to situations where standard procedures might not work. In that case, a non-autonomous robot would wait for human input to continue its task. A robot equipped with a certain level of intelligence could analyse its surroundings and decide on an action without the need for outside input.
Intelligent robots show great potential for use in hazardous or inaccessible environments. Robots could be sent to the deep sea or planetary surfaces to create maps, obtain samples, and build, inspect, and maintain infrastructures. Search and rescue robots could enter hazardous and unstable areas such as earthquake sites to search for survivors and provide useful footage, reducing risks posed to human rescue teams.
Robot autonomy is evaluated using field tests in different scenarios. For example, in an analogue for similar sites on the Moon and Mars, the researchers found that highly mobile robots exploring lava caves in Tenerife orientated themselves by generating landmarks, planning, simulating, and executing their actions as they mapped the caves. However, they still required help when their return path was blocked, or their sensors malfunctioned. This calls for engineering a next generation of robots that are more robust.
The researchers also propose systems with different robots working together with humans using exoskeletons and teleoperation, which are closer to real situations in which autonomous robots could be deployed. Significant human–robot interaction and high levels of autonomy would be required in these cases.
The Moon, Mars, and Jupiter’s moon Europa are key targets for intelligent robotic exploration. The Moon is our nearest planetary body and presents several options for habitation and exploration. Autonomous robots could potentially be employed to build a base on the Moon, which could be used as a starting point for exploring more distant bodies.
Missions to Mars would likely involve exploration, mapping, and sampling of potential resources and determining sites for future infrastructure. The timescales for communication between Earth and Mars are prohibitively high due to distance, making robot autonomy even more important. The robots employed must operate independently without waiting for remote human intervention.
Scientists suspect there may be a deep ocean under a thick ice layer on Europa that could contain life. For this even more distant mission, after landing a probe and digging through the ice, aquatic robots would be an ideal exploration tool. These aquatic robots would necessarily be autonomous, as human intervention would not be practical in this case.
With this in mind, the researchers developed an autonomous aquatic vehicle for this type of mission that could fit into an ice drill, navigate autonomously, use no energy to dive, search the seafloor, and autonomously dock on return to the probe for data transfer. Autonomy is as crucial for deep-sea robots as in space because communication is difficult and unforeseen challenges such as shifts in currents can occur.
For the foreseeable future, humans will remain a key component of robotic missions in space and the deep sea. For example, humans and robots could share tasks for maintenance of space stations such as the International Space Station, in which humans can either directly control robots through teleoperation or instruct them to perform tasks. In this scenario, robots would not just be a tool, but an assistant to the human.
Hybrid teams could be formed in the future, in which robots and humans act as autonomous partners fulfilling a task. These applications require the development of technologies that provide effective communication between human and robotic team members. This type of human–robot cooperation will also be crucial in developing intelligent robotic solutions in healthcare and industrial settings, where effective and precise communication is essential.
Although robots have the potential for widespread industrial applications, there are several barriers to transferring the results obtained in space and deep-sea exploration to industry. Different levels of knowledge transfer can be required for certain applications, from details on individual software and hardware parts, complete robotic frameworks or systems, to the robotics and AI knowledge needed for a particular use.
A research institute such as DFKI does not deliver products, but rather assists with knowledge transfer into applications, for which resources are normally provided by industrial partners, and occasionally, public bodies. The introduction of AI and robotics technologies in a company takes significant time and resources, and sometimes, also initiates transformative processes in the company itself. Here, a research institute like DFKI can become a strong partner in accompanying this process by providing the relevant technology building-blocks, helping to gain the required knowledge and competence with the company, and providing infrastructure for testing and evaluation.
In standard industrial applications, such as a robotic arm in a factory, operating conditions are highly controlled. However, a fully autonomous robot used in everyday life would need to interact with people in constantly evolving scenarios. It must act independently, learn, solve complex tasks, and react to unpredictable situations, and its proximity to humans means that it must be reliable and safe.
As the human environment places high demands on the physical, mechanical, and electrical hardware and AI software of an intelligent robot system, the next generation of robotic systems developed for everyday use will be semi-autonomous rather than fully autonomous. These robots will have some independence, such as the capacity to navigate around obstacles, with humans controlling complex decision-making and activities.
Semi-autonomous robotic systems present many possibilities within care settings, and researchers have developed adaptive robotic motorised beds with a robotic arm and sensors to adjust bed position depending on need, and intelligent exoskeletons worn by neurology patients for their rehabilitation.
Due to the ever-evolving nature of the field, it is difficult to determine when fully autonomous robots will be available for use in human settings. Several promising robotics systems developed for hazardous situations present high levels of autonomy, and Straube and Kirchner suggest that their autonomy and cooperation could be transferred to applications in the human sphere. More research and development is required to ensure the reliable operation of autonomous robots in complex human environments to eventually realise the dream of robots and humans working together as a team.
That’s all for this episode – thanks for listening. Links to the original research can be found in the shownotes for this episode. And, as always, stay subscribed to Research Pod for more of the latest science.
See you again soon.
Leave a Reply