How artificial intelligence is shaping the future of robotics
Before 2023, or at least it feels that way, many people did not give artificial intelligence much thought. A study conducted in 2017 revealed that only 17% of 1,500 senior business leaders in the United States were familiar with AI. Although some of them had problems seeing how it could impact their respective companies, they all understood that there was potential for it to change business processes.
However, by 2023, the awareness of AI had significantly increased. It was the year chatbots like ChatGPT and text-to-video/image generators like Midjourney became popular, and governments worldwide began to take their risks seriously.
Artificial intelligence in robotics
But what is AI exactly, and why is this happening now? The Oxford Dictionary defines it as "the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages." I.e AI comes in many different forms and shapes. ChatGPT and it's like belongs to an increasing group of AI called generative AI; which refers to a category of artificial intelligence that can generate new content, including text, images, videos, music, or code, based on the patterns it learns from existing data.
In robotics, the use of generative AI is growing, alongside established AI forms and machine learning, which are poised to significantly advance the field.
Although we should be aware of AI's potential harmful effects, it's here to stay. It's all about how we use it as a benefit. This article will focus on ways AI is used in robotics and myths about the technology. But first, why is all this happening right now and not ten or twenty years ago?
AI has been around for 60 years - why is this taking off right now?
The early steps of AI were made in the late 1950’s, led by pioneers like Frank Rosenblatt, who made important contribitions to the developement of the Perceptron, one of the earliest types of artificial neural networks, in 1957. The current revolution in AI is primarily fueled by the exponential growth in computing power, following Moore's Law, and the unprecedented availability of big data from digital interactions.
Frank Rosenblatt was an American psychologist notable in the field of artificial intelligence. He is sometimes called the father of deep learning for his pioneering work on artificial neural networks.
Enhanced computing capabilities, particularly through GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units)s, enable complex algorithms to be processed more efficiently, while the vast accumulation of data provides the necessary training material for machine learning algorithms to improve.
Significant advancements in machine learning techniques, especially deep learning, alongside a culture of open-source collaboration, have accelerated innovation. Additionally, substantial investments from both public and private sectors, coupled with increased accessibility to AI education and a growing skilled workforce, have collectively propelled the rapid advancement and widespread adoption of AI technologies. This convergence of technological, economic, and educational factors marks why the AI revolution is happening now.
Seven ways to use AI in robotics
The development of robots is very costly, mainly because a tremendous effort has to be put into programming them to perform specified tasks. In robotics, various types of AI are employed to enable robots to perceive their environment, make decisions, and perform tasks autonomously or semi-autonomously. The kind of AI used in robotics includes:
Machine Learning (ML), which enables robots to learn from data, improve their performance over time, and adapt to new circumstances without being explicitly programmed for every task.
Deep Learning (DL) is a subset of machine learning that uses neural networks with many layers. It's particularly useful for processing and interpreting complex sensory data like images and sound, which are crucial for tasks such as object recognition, navigation, and obstacle avoidance.
Computer Vision allows robots to interpret and understand the visual world, facilitating tasks like object detection, facial recognition, and scene understanding, which are essential for navigation and interaction with objects and humans.
Natural Language Processing (NLP) enables robots to understand and generate human language, allowing them to follow verbal commands, interact with humans through conversation, and read text or instructions.
Reinforcement Learning (RL) enables robots to learn decision-making through actions within an environment aimed at goal achievement, harnessing trial and error and guided by reward feedback. This approach is highly effective for intricate, interactive tasks and decision-making processes, inherently holding promise for substantially lowering development expenses and thus facilitating the widespread deployment of robots.
Robotic Process Automation (RPA) is often forgotten when discussing robots, as it is not "intelligent" in the adaptive or learning sense. RPA is often included under the AI umbrella for its ability to automate repetitive tasks based on certain rules and inputs.
Generative AI has also a role in AI for robotics. For tasks that involve creating new content or responses based on learned data, such as generating diagnostic reports, designing objects, or predictive modeling.
These technologies are often combined in sophisticated robotics systems to create robots capable of performing a wide range of tasks, from industrial automation and medical surgery to customer service and personal assistance.
Five common misconceptions about AI
Though AI has been around since the late 1950’s, the rapid advancements over the recent years has lead to many misconceptions about AI. It is therefore worthwhile spending some efforts on clarifying misconceptions about Artificial Intelligence (AI), which we think is crucial for understanding its capabilities and limitations.
Here are five widespread myths that, from our perspective, hold particular relevance for the field of robotics:
AI will lead to massive job displacement: A prevailing concern is that AI will automate jobs on a massive scale, leading to widespread unemployment. History, through two centuries of industrial advancement, shows us that technological progress tends to generate new roles and demand for novel skills, thereby creating employment opportunities. While AI and robotics will undoubtedly change the employment landscape and automate specific roles, they also pave the way for new job creation and skill requirements, especially in fields related to AI and robotics management, development and maintenance, as well as ethical supervision.
AI can fully replicate human intelligence: It is a common belief that AI can fully emulate all aspects of human intellect. Though AI surpasses human abilities in certain areas, such as complex calculations, analyzing extensive datasets, and even playing chess, it lacks the emotional depth, creativity, and nuanced comprehension inherent to humans.
AI has Bias-Free Operation: Contrary to the belief in AI's inherent objectivity, AI systems can reflect and even intensify biases from their training data. This is a very important aspect of AI, and the notion of AI's impartiality is thus misguided. An AI's outlook is profoundly influenced by the diversity and quality of its training data. Thus, to achieve AI fairness it is vital to understand that not all datasets are equally suitable, and detect weaknesses in the dataset you want to make use of.
AI can attain consciousness or sentience: Despite recemt advances in machine learning, current AI operates solely on algorithms and data processing, devoid of any form of consciousness, emotional depth, or self-awareness.
AI is a monolithic technology: AI is often mistakenly thought of as ‘one’ technology, yet it actually spans a diverse array of technologies and applications. This includes everything from basic automation tools to sophisticated machine learning systems. It is important to have in mind that different types of AI are suited for different tasks. For example, there's a vast difference between narrow AI (designed for specific tasks) and the theoretical concept of general AI (which would perform any intellectual task that a human can).
The Future of Robotics and AI
Future breakthroughs in robotics, fueled by AI, are anticipated to concentrate on self-governing decision-making, allowing robots to function autonomously in intricate settings. This will be driven by new advancements in reinforcement learning, which represents a true gamechanger in the field of robotics.
Anticipated advancements in robotics, energized by AI, will likely focus on autonomous decision-making, enabling robots to operate independently in complex scenarios. This progress is expected to be propelled by breakthroughs in reinforcement learning, as mentioned above, marking a significant leap forward in robotics.
Improvements in natural language processing and emotional AI aim to enhance human-robot interactions. This will facilitate a smoother integration of robots into daily human activities. The advent of swarm robotics, featuring coordinated robot management, promises to revolutionize areas like agriculture, and search and rescue operations.
Enhancements in computer vision and AI will significantly improve robots' environmental perception and interaction capabilities. Customized AI robots will offer personalized assistance, benefiting sectors like personal care and education. Additionally, there's an ongoing effort to develop ethical and socially responsible robots that resonate with human values and societal norms.
These advancements point towards a future with robotics that are more autonomous, intelligent, and seamlessly integrated with human society and workflows. For robot developers, recent achievements, investments, and developments in AI are making commercially viable robots more affordable to develop, simpler to train, easier to use, and more capable of solving real-world problems.
