February 13 News According to VentureBeat, we have heard enough scientific and technological predictions about how robots will take over monotonous tasks in our daily lives, enabling humans to focus on higher-tech or creative tasks. But the actual situation may be contrary to this: the tasks that are most easily automated are the creative work, and the automation of those repetitive tasks is rather tricky. Surprisingly: A field that is easy to automate art In the early 21st century, the “The Painting Fool†painting program at the University of London created works of art. Most of them entered the gallery and were exhibited with human works. Neural networks like DeepStyle and Prisma use convolutional neural networks to imitate specific artist styles to stylize photos. Logo generation systems such as Withoomph, Tailor Brands, and Logojoy can generate logo designs based on keywords semi-automatically or fully automatically. The researchers also applied this process to music. Melomics is a system that creates and plays music based on lifestyle and activity themes. IBM and artists joined forces to combine a large amount of music, lyrics and sentiment analysis to help "Watson" synthesize music. In this way, natural human creativity is no longer necessary to create beauty. research A core foundation of science is repetitiveness. Pipetting operations are monotonous and labor-intensive tasks in many laboratories, and companies such as OpenTrons are committed to using automated pipetting devices to help scientists save time and money. The creation of Arcturus and BioRealize allows scientists to remotely manipulate genetic engineering experiments, greatly reducing errors and experimental time. Other startups, such as Emerald Therapeutics and Transcriptic, hope to move the research process to the cloud and use remote robot systems to automate experiments. In addition to automated manual laboratory work, current machines are also automating scientific discovery and understanding. Cornell University's Nutonian can create data models without any presupposition. Researchers from Cambridge, Asbury, and Manchester created a similar self-help science algorithm, calling it the first machine in history that could independently discover new scientific knowledge. As scientific research increases, scientists may increasingly rely on automation systems. legal Engaging in the legal profession requires years of study and understanding of laws, cases, and other case law. The latest advances in AI make it possible to automate these tasks. So far, case law, contract law and defense have been automated. DoNotPay's robotic lawyers can even help users with traffic tickets. ROSS Intelligence uses AI to elaborate relevant legal passages and cases to improve the efficiency and quality of legal research. eBrevia uses AI to extract data from contracts to help with related applications such as contract analysis. The extent to which the law can be automated can cause people's misgivings. Since most agencies in the legal industry charge by the hour, it would be interesting to see how lawyers weigh the benefits and responsibilities of automation. Police and security The main function of the human physical security team is to observe and report security incidents. Because of their responsibilities to employers, many human security personnel do not allow interference, which makes their work easier to automate. Security automation to date has focused on enhancing existing security forces. The use of inexpensive means to enhance the security field of vision and hearing range, beyond the scope of traditional fixed cameras. Knightscope and Gamma 2 Robotics create ground-based security robots that provide power deterrence through physical presence. These robots are deployed in more and more enterprise parks and shopping malls. Another startup called Nightingale used drones to help customers monitor their operations. In a little while, robots can basically replace the police and security forces. They only need to use artificial forces to resolve violent incidents. Unexpectedly difficult to automate clean The Cleanroom Roomba was one of the first consumer-grade robots introduced to consumers in 2002. Today, 15 years later, the clean-robot industry did not have any substantial innovation to win business success. A company called Intellibot created a large-scale automated floor cleaner called TASKI with very few sales. In addition, Brain Corp has also jointly launched a large floor cleaning machine called "RS26" with the International Cleaning Equipment Corporation. In addition to cleaning floors, robots have also been introduced into cleaning operations that are too dangerous for humans. The founding company Ecoppia uses robots to clean positively charged panels, and Kurion's nuclear pollution cleaning robot, which is specialized in radioactive waste management, is also selected to operate in Fukushima. Clothing and Textile Textile manufacturing is one of the first industries to achieve automation, but full automation is still difficult to achieve. Robots are good at handling solid objects that do not deform, but the cutting, stretching, and compression of textiles make robots helpless. Today's robots can perform dyeing, cutting, and complex pattern designs, but the actual work of sewing, adding lace, and folding still requires dexterous manual work. Few companies have made significant progress in advanced sewing. Softwear Automation is one of them. It uses high-speed cameras to track textiles and let robots work on curved fabrics. Nike is also using automation to replace the traditional shoe assembly process. The top of the FLYKNIT shoes is a continuous thread weave rather than the assembly of multiple parts. There is also little progress in automation in apparel care. Although the machine can be washed and dried, the loading, removal and folding processes are still repetitive and laborious tasks without ideal automation solutions. The company Laundroid and Foldimate developed a laundry folding robot, but it has been criticized for its large size and small functionality. agriculture The use of automation to replace manual crop harvesting has so far been difficult. Some crops are fragile and fragile, and computer vision is not easy to identify crops. But despite these challenges, many companies are still committed to fully automating agriculture. Harvest Automation created a robot called "HV-100" that can almost automate the entire growth cycle of a crop. There are also a series of startup companies dedicated to creating weeding and loose soil robots. This includes Naio Technologies, which manufactures weeding robots for the vineyard. In the field of fruit picking, companies such as FF Robotics, Abundant Robotics, and Vision Robotics have been working to provide robotic fruit picking solutions. Although their products have not yet been popularized, their efforts are laying the foundation for a more automated world. For example, they use a suction device to pick up the fruit, avoiding many of the drawbacks of holding the arm. Looking to the future From the several topics outlined above, we can roughly predict the future of the robotics industry. Today, robotics technology has made significant progress in automating repetitive tasks that require less than superb technology. Tasks involving large amounts of data and information processing can also be multiplied by automation. On the other hand, tasks that require complex operations are still difficult to automate. These tasks require more technical support, including deeper decision-making and information integration capabilities that robots can provide us. Robots have changed many major industries, and fewer and fewer things need to be done manually. But overall, a fully automated world is still out of sight.