Embedded Files
The 10 paths planned with intention estimation of target ships with random initializations.
Long-range path planning for ships in complex and dynamic environments
Long-range path planning for ships in complex and dynamic environments
Used probabilistic roadmap and modified Dijkstra's algorithm with steering cost to construct destination trees. Each path on the destination tree is the optimal path on the roadmap.
Used cosine distance between the target ship's current motion and the path on destination trees to measure the intention score.
Developed risk-aware A* algorithm to find a path with low risk of encountering other ships in the long-range motion. The result showed a reduction of over 60 percent in the duration of encountering at least three ships compared to the algorithm without intention estimation.
Design of Intelligent Agent
Design of Intelligent Agent
An intelligent agent controlled by raspberry pi. The chassis is manufactured by 3D printing.
The agent can detect objects with the local camera and report its real-time location to the laptop.
RaceOn Self-Driving Car
RaceOn Self-Driving Car
A self-driving car controlled by raspberry pi.
The control algorithm is based on computer vision and PID controller.
Lane Detection
Lane Detection
A lane detection algorithm based on Hough Transformation.
Image Perspective Changing
Image Perspective Changing
A view transformation algorithm based on camera pin-hole model. By detecting the corners on a checker board and applying a gradient based optimizing algorithm, an image can be transformed to the top view.
Page updated
Google Sites
Report abuse