INTERACTION Dataset: An INTERnational, Adversarial and Cooperative moTION Dataset in Interactive Driving Scenarios with Semantic Maps (1910.03088v1)

Abstract: Behavior-related research areas such as motion prediction/planning, representation/imitation learning, behavior modeling/generation, and algorithm testing, require support from high-quality motion datasets containing interactive driving scenarios with different driving cultures. In this paper, we present an INTERnational, Adversarial and Cooperative moTION dataset (INTERACTION dataset) in interactive driving scenarios with semantic maps. Five features of the dataset are highlighted. 1) The interactive driving scenarios are diverse, including urban/highway/ramp merging and lane changes, roundabouts with yield/stop signs, signalized intersections, intersections with one/two/all-way stops, etc. 2) Motion data from different countries and different continents are collected so that driving preferences and styles in different cultures are naturally included. 3) The driving behavior is highly interactive and complex with adversarial and cooperative motions of various traffic participants. Highly complex behavior such as negotiations, aggressive/irrational decisions and traffic rule violations are densely contained in the dataset, while regular behavior can also be found from cautious car-following, stop, left/right/U-turn to rational lane-change and cycling and pedestrian crossing, etc. 4) The levels of criticality span wide, from regular safe operations to dangerous, near-collision maneuvers. Real collision, although relatively slight, is also included. 5) Maps with complete semantic information are provided with physical layers, reference lines, lanelet connections and traffic rules. The data is recorded from drones and traffic cameras. Statistics of the dataset in terms of number of entities and interaction density are also provided, along with some utilization examples in a variety of behavior-related research areas. The dataset can be downloaded via https://interaction-dataset.com.

• The paper introduces the INTERACTION dataset, providing a wide range of interactive driving scenarios for improved motion prediction and behavior modeling.
• It leverages semantic maps and data from multiple countries to capture critical adversarial and cooperative driving behaviors.
• The dataset enables rigorous testing of autonomous driving algorithms and offers insights into cross-cultural variations in road user behavior.

Overview of the INTERACTION Dataset Paper

This essay provides an overview of the paper titled "INTERACTION Dataset: An International, Adversarial and Cooperative Motion Dataset in Interactive Driving Scenarios with Semantic Maps." The paper introduces a dataset that is particularly valuable for advancing research in autonomous driving technologies, focusing on the crucial role of interactive motion datasets.

Key Features of the Dataset

The INTERACTION dataset is designed to address several critical aspects necessary for the development of autonomous vehicles, which are not fully covered by existing datasets like NGSIM and highD. Key features of the dataset include:

1. Diverse Interactive Scenarios: The dataset encompasses a wide range of scenarios, such as urban and highway merging, roundabouts, and various types of intersections. This diversity allows for a comprehensive analysis of motion prediction and behavior modeling.
2. International Driving Styles: By incorporating motion data from several countries across different continents, the dataset reflects various driving behaviors and preferences. This diversity supports the paper of how cultural differences impact driving styles and informs the development of adaptable autonomous driving algorithms.
3. Complex and Critical Behaviors: The dataset includes complex interactions that involve adversarial and cooperative driving behaviors, such as negotiations and traffic rule violations. It also spans a range of critical situations, from safe operations to near-collision scenarios, providing a rich resource for evaluating algorithms under challenging conditions.
4. Semantic Maps: Detailed maps with semantic layers, including physical aspects, reference lines, and traffic rules, are provided. These maps are critical for motion planning and prediction research, enabling researchers to utilize contextual information in their analyses.
5. Complete Interaction Data from Bird's-Eye View: Data is collected using drones and traffic cameras, ensuring a complete view of all interacting entities, which is often a limitation in datasets derived from onboard sensors.

Implications for Autonomous Driving Research

The INTERACTION dataset has several implications for autonomous driving research:

• Future Prediction: It serves as a critical resource for improving models designed to predict the future states of road users. The dataset supports both traditional and novel prediction methodologies, enhancing the accuracy and interpretability of predictions in complex scenarios.
• Imitation Learning: The detailed interaction records allow for the development of algorithms that can mimic human driving behaviors in diverse environments. This is essential for creating autonomous systems that can drive naturally and safely alongside human drivers.
• Algorithm Testing and Validation: The dataset can be used to rigorously test and validate decision-making and motion planning algorithms under real-world conditions that include both cooperative and adversarial maneuvers.
• Cultural Adaptation: With data from multiple countries, the dataset offers insights into how autonomous vehicles can be tailored to accommodate different driving cultures, potentially leading to more universally effective systems.

Future Research Directions

The availability of such a comprehensive dataset opens up several avenues for future research. The focus could be on enhancing multi-agent interaction models, exploring cross-cultural variations in driving behavior, and refining real-time motion prediction systems. Moreover, as autonomous vehicles move closer to widespread adoption, datasets like INTERACTION will be instrumental in ensuring that algorithms are robust, reliable, and safe across a multitude of driving scenarios worldwide.

In conclusion, the INTERACTION dataset represents a significant contribution to the field of autonomous driving by providing a resource that captures the nuances of real-world interactions in diverse settings. It supports a wide array of research areas and sets the stage for developing more sophisticated and adaptable autonomous driving technologies.