The past few years have seen active development in autonomous driving by organisations and academia. One of the standard practices in autonomous driving is developing and validating prototypes of driving in simulators. The researchers worldwide have been developing these simulators to support the training and development of such selfless driving systems.
Let’s take a look at the top 8 autonomous driving open-source projects one must try their hands-on.
(The list is in no particular order)
1| Carla
About: Carla is an open-source simulator for autonomous driving research. It has been developed to encourage development, training as well as validation of autonomous urban driving systems. In addition to the open-source code and protocols, this simulator provides open digital assets, such as urban layouts, buildings and vehicles.
The simulation platform promotes flexible specification of sensor suites. Carla can be used to study the performance of three approaches to autonomous driving — modular pipeline, an end-to-end model trained via imitation learning, an end-to-end model trained via reinforcement learning (RL). Carla’s features include scalability via a server multi-client architecture, autonomous driving sensor suite, flexible API, fast simulation for planning and control, maps generation, traffic scenarios simulation, ROS integration, and autonomous driving baselines.
Know more here.
2| SUMMIT
About: SUMMIT or Simulator for Urban Driving in Massive Mixed Traffic is a high-fidelity simulator that promotes the advancement and testing of crowd-driving algorithms. It simulates unregulated and dense urban traffic for heterogeneous agents regardless of the worldwide locations that OpenStreetMap supports.
The SUMMIT simulator is built as an extension of CARLA and inherits the physics and visual realism for autonomous driving simulation. It supports a wide range of applications, including perception, vehicle control and planning, and end-to-end learning.
Know more here.
3| Flow
About: Flow is an open-source computational framework for deep RL and control experiments for traffic microsimulation. Developed by the members of the Mobile Sensing Lab at UC Berkeley, Flow is basically a deep reinforcement learning (RL) framework for mixed autonomy traffic.
This simulator is a traffic control benchmarking framework that gives a set of traffic control scenarios as benchmarks, tools for designing custom traffic scenarios, and integration with deep reinforcement learning and traffic microsimulation libraries.
Know more here.
4| PGDrive
About: PGDrive is an open-ended and highly configurable driving simulator that integrates the key feature of the procedural generation (PG). The simulator defines multiple basic roadblocks such as ramp, fork, and roundabout with configurable settings and a range of diverse maps can be assembled from those blocks with procedural generation, which is further turned into interactive environments.
PGDrive is built upon the Panda3d and Bullet engine with optimised system design. The simulator can gain up to 500 simulation steps per second while running a single instance on PC.
Know more here.
5| Deepdrive
About: Deepdrive is an open simulation platform built to accelerate progress and increase transparency in self-driving. The features of Deepdrive include support for Linux and Windows, interface through the Gym API using a reward function based on speed, safety, legality, and comfort, pre-trained example agent, training code, and dataset to get started building AI models. It is also enabled with up to eight cameras and a dataset of around 100GB and 8.2 hours.
Know more here.
6| AirSim
About: developed by Microsoft, AirSim is an open-source, cross-platform simulation platform for autonomous systems. Built on Unreal Engine, AirSim supports software-in-the-loop simulation with popular flight controllers and hardware-in-loop with PX4 for physically and visually realistic simulations.
AirSim is developed as an Unreal plugin which can be dropped into any unreal environment. The developers at Microsoft developed AirSim as a platform for researchers in AI and to experiment with deep learning, computer vision (CV) and reinforcement learning (RL) algorithms for driverless vehicles.
Know more here.
7| LGSVL Simulator
About: LGSVL Simulator is an open-source autonomous vehicle simulator developed by LG Electronics America R&D Centre. It is an HDRP Unity-based multi-robot simulator for autonomous vehicle developers that provide an out-of-the-box solution that can meet developers’ needs to focus on testing their autonomous vehicle algorithms.
The simulator currently has integration with TierIV’s Autoware and Baidu’s Apollo 5.0 and Apollo 3.0 platforms, can generate HD maps, and can be immediately used to test and validate a whole system.
Know more here.
8| Gym-Duckietown
About: Gym-Duckietown is a simulator for the Duckietown Universe. It is written in pure Python/OpenGL (Pyglet). The simulator works by placing RL agents inside of an instance of a Duckietown: a loop of roads with turns, intersections, obstacles, Duckie pedestrians, and other Duckiebots.
Duckietown is a fully-functioning autonomous driving simulator that can train and test machine Learning, Reinforcement Learning, Imitation Learning, or even classical robotics algorithms.
Know more here.
