As a part of my MSc thesis, I have evaluated the ability of GP to evolve successful control program that can help an autonomous robot to aviod static obstacles through its navigation toward a target position. However, the obstacle avoidance behavior has become a classical problem that does not have a wide interest in the field of automatic programming. Therefore, I am looking for a new challenge that includes training a robot to do more significant navigation tasks based on general solutions gradually evolved be GP.
I think you can research these issues regarding the use of multiple mobile robots:
- MAPF - Multi-Agent Path Finding (Genetic programming for conflict-free path planning solvers)
- DES-CTC MMRS - Discrete Event System - Continuous-Time Control Multiple Mobile Robot System (Genetic programming for optimizing a resource allocation supervisor).
You could try exploring the possibilities of mobile manipulators (robotic arm on a mobile robot), this would also fit.
Research and studies of various techniques used for the navigation system of robots come up against a lot of problems. Obstacle avoidance and trajectory following are considered basic problems in robotic systems. The purpose of navigation is to navigate through a cluttered environment in search of an optimal path from the starting position to the target position. The navigation of the robot is done by deterministic and non-deterministic (stochastic) algorithm and their hybridization is called evolutionary algorithm, which is also commonly used to solve the navigation problem.
Navigation can be divided into two types: Global and local navigation. For global navigation type, prior knowledge of the environment must be known, which is also called offline mode for path planning, for example: A* algorithm (Liu & Gong, Citation2011), potential field method artificial (Gomez, Martinez Santa, & Martinez Sarmiento, Quote 2013). For local navigation, also known as online mode for path planning in which the robot decides its position and orientation and can control its movement using externally equipped sensors, e.g. : an infrared sensor, an ultrasonic sensor, a LASER and a vision sensor (camera) can be used to automatically correct the orientation of the robot via software, other techniques involve NN (Engedy & Horvath, Citation2010), PSO (Ahmadzadeh & Ghanavati, Quote 2012). They are considered to be successful algorithms used by researchers. Local navigation methods are described below.
In general, researchers have mostly used soft versus hard computing techniques, that is, deterministic, non-deterministic, and scalable algorithms for robot navigation and obstacle avoidance.
Nature-inspired algorithms for obstacle avoidance and robot navigation have much more importance in research field and hybridization of deterministic algorithm with non-deterministic algorithm is the next choice for many researchers.
The researchers used these techniques for a static environment and not for a dynamic environment and did not implement them on physical robots, but rather use simulations for their results.
It has been found that for dynamic motion planning, heuristic approaches tend to yield better results.
You don't have to worry about anything. There is a universal controller! There is no need for you to study the control plant and make its mathematical model. There is no need to adjust the controller. Just take and use! https://t.me/universalcontrol
One area of research that may interest you is multi-agent robot navigation, where multiple robots need to navigate in a coordinated manner while avoiding obstacles and reaching a target destination. This involves developing algorithms for task allocation, communication, and coordination between robots. Another area is navigation in uncertain or dynamically changing environments, such as underwater exploration or disaster response scenarios. Additionally, you could explore the use of machine learning techniques, such as reinforcement learning or deep learning, to train robots to navigate complex environments. These are all active areas of research that could benefit from the use of evolutionary algorithms like GP to evolve control programs for robots.
One area of research that may interest you is multi-agent robot navigation, where multiple robots need to navigate in a coordinated manner while avoiding obstacles and reaching the target destination. This includes developing algorithms for task assignment, communication, and coordination between robots. Another area is navigation in uncertain or dynamically changing environments, such as underwater exploration or disaster response scenarios. In addition, you can explore the use of machine learning techniques, such as reinforcement learning or deep learning, to train robots to navigate complex environments. These are all active research areas that could benefit from using evolutionary algorithms such as GP to develop control programs for robots.
- Badges
- Science topic