Two years ago, Stanford Artificial Intelligence Lab Director Fei-Fei Li and her team launched the global platforms RoboTurk and Surreal to crowd-source high-quality task demonstration data that could help researchers working in robotic manipulation. Now, the wait is over. The Stanford Vision and Learning Lab announced this week that the RoboTurk Real Roboto Dataset is available as a free download. The crowdsourcing produced 111.25 hours of video from 54 non-expert demonstrators to build "one of the largest, richest, and most diverse robot manipulation datasets ever collected using human creativity and dexterity." Participants used smartphones and browsers to access the original RoboTurk crowdsourcing platform, where they could remotely control robot simulations in real time.

In our initial publication, we used RoboTurk to collect a large dataset using robot manipulation tasks developed using MuJoCo and robosuite. However, there are several interesting tasks that cannot be modeled in simulation, and we did not want to restrict ourselves to those that could. Thus, we extended RoboTurk to enable data collection with real robot arms, and used it to collect the largest robot manipulation dataset collected via teleoperation. The dataset consists of RGB images from a front-facing RGB camera (which is also the teleoperator video stream view) at 30Hz, RGB and Depth images from a top-down Kinectv2 sensor also at 30Hz, and robot sensor readings at 100Hz. We collected our dataset using 54 different participants over the course of 1 week.

Imitation learning has allowed recent advances in learning robotic manipulation tasks but has been limited due to the scarcity of high quality demonstrations to learn from. RoboTurk is a system that help solve this problem by enabling the rapid crowdsourcing of high-quality demonstrations. This allows the creation of large datasets for manipulation tasks that we show improves the quality of imitation learning policies. We have recently extended the RoboTurk platform to work with real robots and presented this work at IROS 2019. Click the link above to learn more.