Software & Datasets
Open research resources developed across GAMMA projects.
View the full historical catalog →
Open Software
Open-source implementations and research tools from GAMMA projects.
MorFiC
Zero-shot cross-morphology quadruped transfer through morphology-aware critic conditioning.
GitHub → 3D PerceptionSonoWorld
Code for generating a 3D audio-visual scene from a single image.
GitHub → 3D Perception & Autonomous DrivingOpenVO
Open-world visual odometry with temporal dynamics awareness for monocular driving video.
GitHub → Robot NavigationMTG
Mapless trajectory generation with traversability coverage for outdoor navigation.
GitHub → Robot NavigationDTG
Diffusion-based trajectory generation for mapless global navigation.
GitHub → Robot NavigationAdaptiveON
Adaptive outdoor local navigation for stable and reliable robot actions.
GitHub → 3D PerceptionET-Former
Efficient triplane deformable attention for monocular 3D semantic scene completion.
GitHub → 3D PerceptionPoCo-CCR
RGB-D indoor place-recognition code for PoCo and CSCPR.
GitHub → Autonomous DrivingBehavior-Driven Traffic Simulator
A SUMO-based traffic simulator incorporating diverse driver behaviors.
GitHub → Autonomous DrivingTrackNPred
Tracking and trajectory-prediction implementations with evaluation tools for dense traffic.
GitHub → Multi-Agent SystemsD-ORCA
A distributed optimal reciprocal collision-avoidance package for multi-robot systems.
Project & code →Datasets
Benchmarks and collected data for navigation, autonomous driving, crowds, and human-centered autonomy.
GND
Multi-modal perception and multi-category traversability data from outdoor campus environments.
Dataset → Autonomous DrivingTRAF
Dense heterogeneous traffic videos across viewpoints, camera motion, time of day, and road-agent types.
Dataset → Autonomous DrivingMETEOR
Dense, heterogeneous, and unstructured traffic scenarios collected in India.
Dataset → Crowds & Human-Centered AutonomyIndia-Walk
Large-scale pedestrian and crowd data captured in dense real-world scenes in India.
Dataset → Autonomous DrivingDAVE
Diverse atomic visual elements with strong representation of vulnerable road users.
Dataset → Crowds & Human-Centered AutonomyGroupWalk
Walking-crowd videos collected across eight real-world locations.
Dataset → Human-Centered AutonomyEmotion-Gait
21-joint gait sequences annotated with happy, sad, angry, and neutral perceived emotions.
Dataset → Human-Centered AutonomyEmotionWalk
16-joint gait sequences annotated with four perceived emotion classes.
Dataset → Human-Centered AutonomyVideoSham
Video manipulations beyond faces, with human and machine analysis.
Dataset →