AI and Robotics

Artificial intelligence (AI) is never far from people's minds in the discussion of robotics. However, robotics and AI are two distinct fields that can and do exist independently. An industrial robot programmed to weld parts on an assembly line does not necessarily need AI. Conversely, an AI system can run on a basic computer without a mechanical arm or sensors. But since 2010, robotics has arguably been the sector where the impact of AI has been the most spectacular and tangible. It is like a showcase of AI’s progress; one that strikes the imagination, arouses fascination and sometimes concern.

In this article, we invite you to explore this fascinating synergy between artificial "body" and "mind".

The natural convergence between robotics and artificial intelligence (AI)

Since the dawn of time, humans have dreamt of creating machines in their own image, endowed with an intelligence of their own. In Ancient times, Homer already sang the praises of divine automatons, those marvels of ingenuity forged by Hephaestus himself. These foundational myths have nourished the imagination for centuries, inspiring pioneers who, from the Middle Ages to the dawn of the Industrial Revolution, sought to breathe life into ever more elaborate mechanical creations.

But it was in 1950 that a decisive turning point occurred. In a landmark article, "Computing Machinery and Intelligence", the visionary Alan Turing asked a question as simple as it was profound: can a machine think? More precisely, can we conceive an artefact whose behaviour would be indistinguishable in every way from that of a human being? This is the famous Turing test, a challenge that would long seem insurmountable.

It wasn’t until the 2010s and the advent of deep learning that Turing's dream to start taking shape. This revolutionary technique, inspired by the very functioning of our brain, opens dizzying perspectives. Thanks to artificial neural networks capable of learning by themselves, machines suddenly became endowed with perception, understanding, and even reasoning. They can recognise a face, transcribe speech, and analyse language with unparalleled finesse.

This is where robotics comes into play. By integrating these AI models into physical systems, we literally give a body to artificial intelligence. Robots cease to be simple automatons and become autonomous entities, capable of:

• Seeing and understanding their environment:  Identifying the objects around them, such as parts to be assembled on a production line, by being equipped with cameras and image recognition algorithms.

• Planning and adapting: Breaking down a complex task into steps and finding the best way to accomplish it, recalculating their plan if an unforeseen event occurs.

• Learning independently: Improving with practice, learning the best gesture to grasp an object, by trying again and again.

• Communicating naturally: Conversing with us in an almost human way by combining speech recognition, speech synthesis and language understanding.

Current applications of AI in robotics

Today, there is a fascinating cross-fertilisation between Robotics and AI. On the one hand, AI allows the creation of more efficient and versatile robots, present in industry, healthcare and services. On the other hand, robotics encourages AI researchers to take up new challenges such as how to interpret complex visuals and how to handle delicate objects. 

Industry

In industry, AI allows robots to be more flexible and adaptable. Let's take the example of a car factory. Assembly robots are now equipped with cameras and image recognition software to verify the conformity of parts, detect defects and adapt to new parts without reprogramming. This is machine learning: the robot learns from experience, like a human.

Cobots, or collaborative robots, are another example. Designed to work alongside workers, they detect human presence thanks to their sensors and AI and adjust their movements accordingly. They also understand voice commands and gestures, making collaboration more natural and intuitive.

Healthcare

In healthcare, robot-assisted surgery is one of the most impressive applications. The surgeon controls precise robotic arms that perform meticulous gestures inside the patient. AI guides these gestures by analysing 3D images in real time and can correct the surgeon's tremors. The result is more precise, less invasive surgery with fewer complications and faster recovery.

Robots also assist caregivers on a daily basis. Autonomous mobile robots transport medicines, samples or medical equipment between departments. Guided by their AI and sensors, they navigate corridors, take lifts and interact with staff via a touch screen, freeing up time for caregivers.

Catering

Some establishments use robots to welcome customers, take orders and serve dishes. Equipped with voice recognition and natural language interfaces, they converse with customers and make recommendations based on the analysis of previous orders.

In the kitchen, robots assist chefs in preparation. Some cook burgers or pizzas autonomously, from preparation to cooking. AI controls each step to ensure consistent quality. Other robots chop vegetables, measure ingredients or assemble plates, assisting cooks in repetitive tasks so they can focus on creativity and quality.

Agriculture

Agriculture greatly benefits from AI and robotics. AI-equipped camera drones monitor crops. They take detailed aerial images, analysed to detect diseases, pests or nutrient deficiencies. Farmers can then take targeted action, reducing the use of pesticides and fertilisers.

On the ground, autonomous robots weed, sow or harvest. Weeding robots use computer vision to distinguish weeds from crop plants and precisely apply small doses of herbicide. For harvesting, picking robots recognise ripe fruits and pick them gently. Some even work at night, an asset for rapid harvests.

AI and robots at the heart of our daily lives

AI is increasingly present in our homes through what is called home automation, or the smart home. This is a set of technologies that allow the control and automation of various home equipment, such as lighting, heating, shutters, or household appliances.

At the heart of these systems, there is often an intelligent voice assistant, such as Amazon's Alexa, Apple's Siri or Google Assistant. Thanks to AI, these assistants are able to understand our voice commands in natural language and respond in a relevant way. We can ask them to turn on the lights, adjust the temperature, give us the weather forecast or launch our favourite playlist, all without moving from the sofa.

But AI is not limited to voice assistants. More and more devices are becoming "smart" thanks to algorithms that allow them to adapt to our habits and preferences:

• Domestic robots: Among the most popular are robot vacuum cleaners. Equipped with sensors and navigation algorithms, they map our interior and clean autonomously. But other robots are being developed to assist us in various daily tasks, from folding laundry to preparing simple meals.

• Educational robots: They are used to teach programming and stimulate students' creativity. Others serve as educational assistants, helping students learn to read, count or practise a foreign language. Thanks to their AI, they can adapt to the level and pace of each student, thus offering personalised support.

• Smart toys: Dolls can have simple conversations with children, while toy robots can interact in a personalised way with each family member. Even our digital leisure activities are influenced, with video games using AI to create more realistic opponents.

Case studies: When AI and robotics revolutionised industries

Siemens: a factory of the future in Amberg

The German giant Siemens has developed a fully automated factory in Amberg, Germany. In this "factory of the future", collaborative robots (or cobots) work hand in hand with human operators to produce electronic components. Thanks to AI, these cobots adapt in real time to production demands and contingencies.

Machine learning algorithms continuously analyse data to optimise processes and predict breakdowns. The result is near-perfect quality (99.9988%), 30% increased productivity and halved costs. A striking example of what Industry 4.0 can bring.

Exotec: the Skypod logistics robot at Renault

French startup Exotec has developed the Skypod, an intelligent mobile robot capable of moving through the aisles of a warehouse to fetch ordered products. At Renault, 191 of these robots have been deployed in the spare parts warehouse. Instead of employees walking kilometres every day, the Skypods bring the parts they need directly to them.

An AI software coordinates this ballet of robots to optimise each movement. Order processing time has thus been reduced from 2 hours to only 20 minutes! This significantly improves the quality of service and Renault's competitiveness in the automotive aftermarket.

Intuitive Surgical: the Da Vinci surgical robot

Intuitive Surgical is the world leader in surgical robotics with its Da Vinci robot. Present in more than 8,600 operating tables, it has already assisted surgeons in more than 7 million procedures. Thanks to its miniaturised instruments and high-definition 3D vision, the Da Vinci allows for extremely precise operations.

With a turnover exceeding 7 billion dollars, Intuitive Surgical proves that medical robotics is also a very buoyant market, at the service of health.

AI and robotics: A detailed overview of major firms

Advanced robotics

• Boston Dynamics: A true pioneer in mobile robotics, Boston Dynamics made a name for itself with Spot, its dog-like quadruped robot. As agile as an animal, Spot is able to climb stairs and open doors. Videos have shown it at work in hostile environments, dangerous construction sites or disaster areas. With Atlas, it has also created a humanoid robot with stunning physical prowess, capable of performing back flips and lifting heavy loads.

• Agility Robotics: In the race to create ever more humanoid robots, Agility Robotics surprised everyone with Digit, a bipedal robot with arms and hands. Capable of lifting packages weighing up to 18 kg and navigating environments designed for humans, Digit seems to come straight out of a science fiction film. But its most promising application is very real: it has been tested by car manufacturer Ford to deliver parcels, carrying them from the truck to the customer's door.

Industrial automation

• ABB Robotics: This young startup is the most ambitious in human-machine collaboration with YuMi, a two-armed robot specifically designed to work alongside workers. YuMi excels in delicate tasks, such as assembling tiny watchmaking components.

• KUKA Robotics: A major player in industrial robotics, the German firm has a whole range of robots. Its flagship, the KR Quantec, is a giant robotic arm capable of lifting up to 300 kg. It has already conquered car factories around the world, where it handles heavy body parts.

AI powering robots

• Nvidia: The graphics processor giant has embarked on a new adventure with Jetson Xavier, a chip no bigger than a credit card. This miniature but super-powerful "brain" allows robots to analyse in real time the images captured by their cameras. With already more than 6,000 applications, such as in recycling centres where robots equipped with Jetson manage to sort household waste (a particularly difficult task, even for a trained human eye!).

The challenges of an increasingly robotised and intelligent world

In this frantic race towards AI and robotics, it is crucial to reflect and install the necessary safeguards. For while these technologies hold immense promise, they also raise unprecedented legal, ethical and societal challenges.

The legal framework

On the legal front, our laws and regulations seem overtaken by the pace of progress. How can we regulate these algorithms that make increasingly important decisions, from granting credit to sentencing? How can we protect our privacy in the face of these robots that collect more and more data about us every day? There is an urgent need to adapt our legislative arsenal to ensure responsible development that respects our fundamental rights.

The ethical framework

But the law is not enough. We also need an ethical compass to navigate these murky waters. Can we entrust AI with life-or-death decisions, whether it's a medical diagnosis or driving an autonomous vehicle? How can we ensure that these systems do not reproduce, or even amplify, our biases and prejudices? In 2016, Microsoft had a bitter experience with Tay, a chatbot that started making racist comments on Twitter. It is time to open a broad societal debate to define together the limits not to be crossed.

The risks of hacking and bugs

On the technical side, vigilance is also required. Despite their prowess, AI systems remain fragile, subject to errors, bugs and hacking. Imagine the consequences of a hacked medical robot, a failing stock market algorithm. The stakes of security and reliability are crucial and require an unprecedented research and standardisation effort.

The impacts on employment

The impact of AI and robotics on employment is also a major concern. According to one study, 20 million industrial jobs could disappear by 2030, replaced by machines. This is a real social shock in prospect. There is an urgent need to invest in training to prepare workers for the jobs of tomorrow. And to rethink our social protection to support those who will be most affected by this transition.

Conclusion

One thing is certain: the convergence of AI and robotics is only in its infancy. In the factories of the future, cobots, these collaborative robots designed to work hand in hand with humans, capable of learning from their mistakes and adapting to our presence, will redefine the very notion of work.

But it is perhaps in our streets that the revolution will be most visible. Autonomous vehicles will transform the way we travel. Safer, and more efficient, they will certainly make traffic flow more smoothly and reduce accidents. The mobility of the future will be intelligent, connected, and perhaps even shared.

To make this vision possible, AI-equipped robots will have to overcome many more hurdles. Knowing how to recognise images, process natural language, and reason about abstract subjects. The challenges are immense, but the progress is just as dizzying. With the advent of deep learning, machines are learning at a breakneck pace, getting closer every day to human capabilities.

This is why the race for AI must not make us forget the ethical issues it raises. The questions are numerous and pressing: how to manage biases in algorithms? How to respect privacy? Who will be responsible in the event of an accident? We must answer them collectively, to ensure that this technology remains at the service of humans and not the other way around.

The impact on employment is another major challenge. While some jobs will inevitably be automated, others will emerge, around the design, maintenance and supervision of these intelligent systems. The transition will be complex and will require an unprecedented effort in training and adaptation. But it also carries hope, that of a society where the most tedious and repetitive tasks will be entrusted to machines, freeing humans for more creative and fulfilling activities.