While knowledge of and familiarity with ChatGPT (an artificial intelligence (AI) chatbot) are quite common, the specific impact of AI on sensors is not. With the increasing use of AI in products and systems, Raj Khattoi, Sr. Director of Sensor Systems and IoT at Infineon Technologies gave SensorTips some answers to specific questions.
SensorTips: With AI, will sensor specifications get tougher requiring increased resolution and/or lower power consumption, or will some parameters be able to be reduced, such as the accuracy?
Khattoi: The shortest answer to this question is what applications engineers working in this field like to say, “it depends” and, in this case, that is exactly true. With the increased use of AI in products and systems, sensors specifications could evolve in very different ways and to understand this, one needs to realize that fundamentally AI is really a complement to the data that is already being generated from these sensors. While some AI applications may benefit from higher resolution sensors to capture more detailed data, others may prioritize lower power consumption to optimize energy efficiency and hence ask for lower resolution sensors. It will also depend on the safety criticality of these applications. Some applications that will need more precise sensors even with the use of AI would be autonomous driving, robotics, and medical applications.
In some applications, the advancements in AI algorithms and processing power may enable more sophisticated data analysis even with reduced data sets, reducing the need for extremely high-resolution sensors and point data clouds. Examples of such sensors could be image processing sensors used in video monitoring and in Natural Language Processing (NLP) systems used in audio assistants for voice recognition. Ultimately, sensor specifications will be determined by the specific needs of AI applications. The advancements in the field of AI will also determine whether or not we will need more or less precise sensors.
SensorTips: With AI, will less sensors be required in aggregated applications or will more sensors be desired?
Khattoi: There will definitely be the need for new types of sensors in all applications and more so in aggregated applications, such as crop planning in vineyards or cattle management. However, the number of sensors in those applications will depend on the application. In other words, in some cases, AI can enable more efficient data processing and analysis, allowing for the aggregation and interpretation of data from fewer sensors. This can lead to cost savings and streamlined operations. On the other hand, AI-driven systems may also create opportunities for more sensors to be deployed in order to capture additional data points and improve the accuracy of AI algorithms.
For example, in cattle monitoring, more sensors could be used to gather data on individual animals’ health, behavior, or environmental conditions. Ultimately, the decision to use fewer or more sensors in aggregated applications will depend on factors such as the specific application requirements, the cost-effectiveness of sensor deployment, and the desired level of data granularity for AI-driven analysis and decision-making.
SensorTips: Are there any specific areas that are already being significantly impacted by AI?
Khattoi: AI has practically had an impact on all fields in the field of sensors and as time progresses, it will have an even bigger impact. One specific area in sensors that is already being significantly impacted by AI is image and vision sensors. AI algorithms, particularly those based on deep learning and computer vision, have revolutionized image processing and analysis. AI-powered image sensors can now perform sophisticated tasks such as object recognition, facial recognition, and image classification in real-time. These sensors can detect and analyze complex visual patterns, allowing for applications like autonomous vehicles, surveillance systems, quality control in manufacturing, security monitoring and even medical imaging. Extensions of the same AI technology using sensors is now being used in novel applications like vineyard management, forest fire detection, and cattle management.
SensorTips: Where would you expect increased future interest in AI+sensors to occur?
Khattoi: I think that there would be a lot of increased future interest in using AI in the field of sensors for driving decarbonization. Here, the combination of AI and environmental sensors is expected to gain traction. These sensors can monitor air quality, pollution levels, water conditions, and weather patterns. AI algorithms can analyze the data from these sensors to provide valuable insights for environmental management and decision-making. The effects of climate change are felt everywhere but have made the life of farmers extremely hard. This is where I can see a growing need of the fusion of AI and sensors for better crop management and agriculture. There are already several wineries in the world that have implemented AI and sensor technologies to enhance their operation. One example is Château Clerc Milon, France. This vineyard in Bordeaux, France, uses AI-powered sensors to collect data on soil moisture, temperature, and vine health. The data is analyzed to optimize irrigation, determine the ideal harvest time, and implement precision viticulture practices.
Roger Grace, president of Roger Grace Associates, a MEMS and Sensors Industry Marketing Consultancy that annually publishes the MEMs report Card, has high expectations for the combination of AI with sensing.
“I consider the AI/Sensors opportunity to be great indeed,” he says. “The availability of low-cost sensors, for example MEMS and printed/fabric, feeds into the basic premise of AI — collecting lots of data — and from the data extracting valuable information to make decisions.”