deeplearning-repo-6

Trajectory prediction for avoidance driving

GPT

Good Prediction of Trajectory

This project is not Open AI's GPT. However, due to the influence of the name, it was produced as GPT (Good Prediction of Trajectory). This project aims to predict the trajectory of objects for avoidance driving.

---

Ball Trajectory

---

Carla Simulator

---

Env.

Ubuntu 22.04 python 3.10.8 tensorflow 2.6.0 numpy 1.23.5 ultralytics supervision 0.3.0 openCV 4.7.0

---

Introduce

Hi! We carried out this project assuming the LIDAR situation as a deep learning project. We first show trajectory predictions for the ball and then use Carla Simulator to predict trajectories for the car.

---

Datasets

If you want to learn, use the trajectory dataset for the ball dataset and the carla dataset.

---

Usage

If you want to predict the trajectory for the ball in real time?

    python3 source/mutiple_prediction.py

If you want to predict trajectories in real time using a Carla simulator?

• If so, look at carla_trajectory_prediction.ipynb

---

Model

The first model.h5 is a predictive model for the trajectory of the ball. And GRU_model.h5 is a model for better ball trajectories. The carla_GRU_model.h5 model is a predictive model for the trajectory of a car.

---

Credits

• Thanks to PinkWink