KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#developer-tools #claude-code #codex #typescript #ai-agents #dsh #dsh-plugin #deepseek-harness #ai-agent #mcp #agent-skills #macos

intelligent-self-driving-car

★ 38 Open GitHub ↗

High speed autonomous vehicle navigation and lane change in dense traffic scenario using custom SUMO gym Environment and Reinforcement Learning algorithm

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

#### RL-Agent (red car) learned deep navigation with intelligent lane change
![navigation](https://raw.githubusercontent.com/A-D-Alamdari/intelligent-self-driving-car/main/images/results/trimed-value.gif)

# Drive Through Deep - An Intellignet Autonomous agent driving in a densely highway scenario.

##### A reinforcement learning framework of a self-driving car to learn lateral (lane change decision) and longitudinal control (acceleration/deceleration) while navigating through deep traffic.

Taks Objectives:
- Make an autonomous agent to learn lane change and longitudinal control to navigate through deep traffic while maintaining better safety and higher traffic flow
- Provide a simple openAI like gym environment to work with reinforcement learning (RL) aglgorithm for lane change and highway driving scenario using Simulation of Urban Mobility (SUMO) simulator.

## Approach
1. Create a custom gym environment using SUMO
2. Define highway driving and lane-changing problem as Markov decision process (MDP)

### Installation
---
#### Install Latest version of [SUMO](https://sumo.dlr.de/docs/Downloads.php)
#### Install custom SUMO Environment
``` Ruby
git clone git@github.com:lokesh-c-das/intelligent-self-driving-car.git
cd SUMO-RL-ENVIRONMENT
cd gym_sumo
pip install -e .
```
## Highway driving & lane changing problem formulation
We formulate the high driving and lane changing problem as a Markov decision process (MDP), where
### State Space:
The state space contains 19 different states of current environment conditions; from ego vehicle (aka RL agent) and its 3 leaders from inlane, target-left lane, target-right lane, and lanes average speed, and lanes density
### Action Space
5 discreate action values that map
+ 0 --> Keep current lane and speed
+ 1 --> Change lane left
+ 2 --> Change lane right
+ 3 --> Accelerate (Constant amount)
+ 4 --> Decelerate (Constant Amount)
### Reward
The reward function consists of a efficiency reward, lane change penalty reward, and a safety reward

### Efficiency Reward:

 <img src="./images/efficiency_reward.png" width="400" height="80" />

### Safety Reward

 <img src="./images/safety.png" width="200" height="45"/>

### Lane Change Penalty

 <img src="./images/lc_penalty.png" width="200" height="35"/>

### Results

#### Training Reward and Loss
![Loss](https://raw.githubusercontent.com/A-D-Alamdari/intelligent-self-driving-car/main/images/results/Loss_train-2.png)![Reward](https://raw.githubusercontent.com/A-D-Alamdari/intelligent-self-driving-car/main/images/results/Reward_Train-2.png)

intelligent-self-driving-car

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README