The discussion of disadvantages of recent robot learning methods and possible solutions

Datong High School, Shanghai, 200400, China

^* Corresponding author: gongshuisantsuye@gmail.com

Abstract

Nowadays, robots have been greatly developed by scientists and robotic experts, especially these operating and coding methods. They are now an important part of the whole society, providing much productivity for human being. However, there are also some obstacles that have not been fully overcome, including unsatisfying database quality, ineffectiveness of AI self-learning, etc. These problems can be categorized into two types: the kinematic problems and AI’s problems. The former includes difficulties in calculations and singularity points, and the latter includes AI’s lack of some certain types of abilities. This essay will focus on some of these obstacles that most robotics industry practitioner and experts are facing, and figuring whether there are some possible solutions. This essay will firstly introduce a strawberry harvester robot project that I’ve participated in and been responsible for its coding works and operating method construction, discuss some of its disadvantages, and then lead to the discussion of common problems across the industry. These problems can be categorized into two types: the kinematic problems and AI’s problems. The former includes difficulties in calculations and singularity points, and the latter includes AI’s lack of some certain types of abilities.