Artificial Intelligence Detection and Planning Tasks New Frame Developed
A new AI framework developed to solve computer vision and decision-making problems, develops planning capabilities by maintaining uncertainty. The research aims to exceed the limitations of traditional methods and carries a wide range of potential from robotic applications to complex visual problem solving.
The basic features of this framework called Neuro-Soft-Symbolic:
- Diferable design: All system can be trained in end-to-end, detection errors can be re-strained directly to the planning module
- Unfortunate management: The uncertainty in visual detection is protected and is included in the planning process, making solid decisions
- Sembolic logic integration: Logical rules are produced directly to the system by embroidering the current motion sequences
- Back notification relationship: Planning tasks help the detection model perform better
Visual data in traditional methods is converted into separate symbolic facts, valuable information and uncertainty disappears in this process. Furthermore, the error signals at the task level do not reach the detection component, the system is limited to the learning capacity of the general.
This new approach allows you to establish a tight connection between detection and planning by solving this problem. As a result, it is possible to develop more accurate and reliable autonomous systems.