In 2010, Rajnikanth-starrer Enthiran gave us a glimpse into the life of ‘Chitti’, an artificial intelligence (AI)-powered humanoid robot, who could fight, clean and even cook but was not programmed to eat or taste food.
Cut to 2022, robot chefs are for real, overshadowing ‘Chitti’. Researchers from the University of Cambridge’s Bio-Inspired Robotics Laboratory said on Wednesday that their robot chef could assess how salty a dish is at different stages of the chewing process, just as we humans do. The researchers worked in collaboration with domestic appliances manufacturer Beko to achieve this goal.
For humans, the sense of taste has evolved over millions of years and remains very subjective—some may like a salty preparation, some spicy, and others less spicy, but how will a robot chef decipher the complexities of something so subjective as taste?
The researchers acknowledge that preparing and cooking food is a challenge “since it must deal with complex problems in robot manipulation, computer vision, sensing and human-robot interaction, to produce consistent end products”. They said robot chefs could personalize tastes by imitating the human processes of chewing and tasting. “When we taste, the process of chewing also provides continuous feedback to our brains,” said co-author Arsen Abdulali from the university’s department of engineering, adding that the current method of electronic testing is inadequate as they are time-consuming and take a single reading from a homogenized sample.
In June 2020, researchers from the same university did use machine learning to train a robot to understand the subjectivity of taste by the taste of omelettes. This time around, the researchers attached a conductance probe with a salinity sensor to a robot arm (conductivity is the movement of ions. Hence, the salinity increases with ion concentration, ion mobility and the ionic charge). They prepared scrambled eggs and tomatoes but added salt and tomatoes in varying quantities for each dish. The robot was able to “taste” the preparations using the probe.
To mirror the change in texture caused by limited chewing, unlike humans who continuously chew to taste, the team then put the egg mixture in a blender and had the robot test the dish again. The different readings at different points of ‘chewing’ produced taste maps for each dish. According to the researchers, the results showed a “significant improvement” in the ability of robots to assess saltiness over other electronic tasting methods.
Likewise, IBM Research had been developing Hypertaste —an AI-assisted tongue mimicking the way men taste things, according to a 5 July 2019 blog by IBM Research staff member Patrick Ruch. The idea was to cater to a range of industrial and scientific users who need to identify liquids swiftly and reliably without access to high-end laboratories. ‘Hypertaste’ uses electrochemical sensors comprised of pairs of polymer coating-covered electrodes that release a voltage signal—the combined voltage signals of all pairs of electrodes represent the liquid’s fingerprint. Key to the functioning of the electrochemical sensors is polymer coatings covering each electrode. Robot chefs have been in existence for almost a decade but are now gaining traction in India since Indians love to eat fresh food.
But robot chefs are not cheap—domestic variants start at ₹50,000, and commercial models can cost up to ₹45 lakh. For instance, Bengaluru-based Euphotic Labs’ smart robot, Nosh, can cook a variety of simple Indian dishes such as upma, poha, halwa, pasta, khichdi, and fried rice. Users just have to feed the ingredients in separate compartments and select the dish from its mobile app that offers over 200 recipes.
Nosh uses a “recipe engine” to autonomously decide to cook a food item or ingredient in a specific way. It also uses a vision algorithm that monitors the food while it is cooking and takes the next steps in the cooking process accordingly, just like humans do. The first prototype for Nosh was developed in 2018, and a commercially ready model will be launched during Diwali this year. Nosh is compact in design and can be deployed at home.
Yatin Varachhia, co-founder and head of product at Euphotic Labs that created Nosh, explains that “Nosh has the intelligence to detect if the intermediate stage of cooking is done properly as per recipes, i.e., onion turned golden brown or is translucent, the curry released oil, rava turned brown, etc.” According to him, “so far 500+ people have tasted the food and rated food taste 4.4+ on average. Also, we have 650+ pre-orders and a long waiting list for people who want the product in their life. The pre-order cost was ₹40,000. Now, the price for India will be ₹49,999, and the US will be about $999. We are not taking the pre-orders now.”
Another Chennai-based startup called Robochef has developed a robotic cooking machine with in-built internet of things (IoT) sensors. It is a lot bigger and is meant for restaurants and cloud kitchens. Its commercial variants have a capacity of 37 litres and can cook 100 servings at a time. Created by Chennai-based engineer Saravanan Sundaramoorthy, Robochef is programmed to clean and chop vegetables, grind spices and cook using the right proportions.
Some of the robots are developed with specific fast-food items in mind. That said, all these robo chefs are yet to develop taste buds. The University of Cambridge robot chef promises to solve this issue too.
