Capital assets are very expensive and they essentially need safety in industry. They are expensive and hence require regular inspection. They are very large structures at distant sites. So, there is need to access them. They are also located in extremely hazardous environment. Such industries include oil and gas, nuclear, wind mills, transportation and railways which are massive and require regular inspection. Humans cannot do in-service inspection due to high temperatures and inaccessibility and robots are certainly a better option. In non-destructive testing, visualization of data by expert and highly knowledgeable inspectors can be augmented with better tools to enable quick decisions. These robots are especially useful for visual inspection and surveillance. Robots are also helpful in assessment of oil and gas tanks for leakage, corrosion and waste deposition. Climbing up robots is instrumental in assessing the integrity of structures from nuclear plants to multistory buildings. Lot of research is being carried out to build the mobile robots which can access safely critical infrastructure, ensure its integrity, reduce its inspection and maintenance cost, reduce outage and turnaround time and if possible perform non-destructive testing which will save more cost. Of course, ultimate aim is to protect divers and operators who have to operate in these dangerous environments. There are a lot of fatalities among divers who try to inspect these structures in the sea and depths.
Machine learning systems must be really transparent. They must be completely free from bias. Success of these robotic machines depends on the training which takes long time. Sometimes, wrong correlations and inferences are made by these machines whose results may be fatal. Other type of problem is bias, stereotypic unfair determinations and inconsistency. UK report on artificial intelligence identifies that implications arise from the volume of huge data, where it is generated and propensity to find new uses for it. A super organism may be created in longer term by connecting robots, computers and every modality which may be misused by some malicious person. People fear that machines are controlling aircrafts to life support systems and if they get hold of whole knowledge and encyclopedia, they may not get out of the hand of humans. It would certainly be catastrophic. Mitigation of these risks is to build AI algorithms and keep them under observation to find bias. Then, there must be research to make deep Artificial neural network more transparent. Governments might be building sensitive AI which is dealing with our health records etc. There is regulation being introduced by EU to protect use of data. So, an EU citizen may demand that its data may be deleted or forgotten. However, risks and benefits of these robotic machines are still debatable.
|