Intelligent RISK-SENSING ROBOTS ROUND THE CORNER
Robots can be of great help to humans. But the lack of risk perception among robots has worried scientists about unwanted outcomes from robotic actions. Humans analyze risk all the time…while walking, driving, swimming, financial planning, conducting surgeries, and even in sports. But if assistance from autonomous robots is needed in different human spheres of activities, this lack of risk perception among robots could be crippling instead of any assistance. Robots can’t take autonomous decisions when risk is involved. Imagine the consequences when a robot driving a car perceives a stop sign as a speed limit sign and keeps moving on – a disaster waiting!
A world with risk-analyzing autonomous robots could make a world of difference in assisting humans, for instance, in disaster recovery in which autonomous robots could be of immense help. Now, researchers from Lehigh University are on the way to set that right. They are quantifying the sources of uncertainty among the robots. They are analyzing a range of perception modules in machine learning models that use visual sensing to better understand how the models perceive the environment surrounding them.
They are studying and analyzing the inputoutput relationship in a robot and compute the uncertainty. Once the ambiguity is quantified, they will use risk-measures to set decision-making purposes in the robot. This will prepare the robots to assess various risk situations and make quick decisions about how to avoid the risk or manage it. The team is looking at a team of risk-assessing robots, capable of assessing risk and taking safe decisions on their next course of action, and who can also communicate with each other effectively. These robots are being planned to perceive and understand human actions and infer how they can best assist them.
A NEW FISH MAY FIND ITS WAY INTO OUR PLATES
Marine aquaculture has a new potential in farming a fish species that scientists have identified as worthy of making its way into plates and on food counters in the next few years. Scientists from the University of Gothenburg in Sweden are working on developing in-land breeding and high-protein feeding techniques for wolf-fish (Anarhichas lupus), which has been identified as an ideal species for consumption to expand the seafood culinary diversity. Dr Ida Hedén, from the Department of Biological and Environmental Sciences at the University of Gothenburg, feels growing only one type of fish isn’t sustainable in terms of the diversity of the market and in terms of food security.
The wolf-fish is naturally found in confined spaces close to the sea bed. Fish breeders could consider this as an added option to expand seafood variety. It is a fish with firm, white flesh that works well on the grill and with boiled potatoes and egg sauce, points out Hedén, who has been a chef in the past. As wolf-fish feast on a high-protein diet, Hedén is working on extracting shrimp and herring protein from processed water from the fish processing industry as feed to reduce dependence on fish meal as wolf-fish’s diet. High-protein feed extracts from waste products reintroduced into food production in a circular approach can be an ideal way of recycling the value from waste while saving on plant-based protein which could conflict with human consumption. But the wolf-fish is slower-growing than the salmon. Hedén is working on developing faster-growing strains of wolf-fish to tap the potential to farm wolffish in land-based fish farms, with tanks on land offering good opportunities to control and create a good growing environment for the fish, and reduce the conflict with other coastal interests.