Entities

Things that come up in many notes. Each page lists the notes that mention it, so these are the hubs of the graph. Generated by bun run diary index.

Datasets

3RScan: RGB-D scans of 478 indoor spaces, each rescanned over time as objects moved, 1,482 scans in all (Wald and others, 2019). (8 links)
Matterport3D and HM3D: Matterport3D has RGB-D panoramas of 90 building-scale scenes (Chang and others, 2017). (12 links)
Replica: Photorealistic reconstructions of 18 indoor scenes from Meta Reality Labs (Straub and others, 2019). (12 links)
ScanNet: RGB-D video of indoor rooms with 3D reconstructions and semantic labels, about 1,500 scans (Dai and others, 2017). (12 links)
uHumans2: Simulated indoor and outdoor scenes with moving people, from MIT SPARK, released with the 3D dynamic scene graph work. (9 links)

BlueROV2: Blue Robotics’ low-cost remotely operated underwater vehicle. (6 links)
Clearpath Husky: Clearpath’s larger four-wheeled outdoor robot. (6 links)
Clearpath Jackal: A small four-wheeled outdoor robot from Clearpath, common in field tests of mapping and navigation. (7 links)
Franka arms: Franka Robotics’ seven-joint arms (Panda, FR3), the usual arm in manipulation and policy learning papers. (6 links)
Intel RealSense: Intel’s RGB-D cameras, such as the D435 and D455. (10 links)
NVIDIA Jetson: NVIDIA’s embedded GPU computers (Xavier NX, Orin Nano, Orin NX, Orin AGX), the usual onboard computer on small robots. (16 links)
Spot: Boston Dynamics’ quadruped. (13 links)
Unitree robots: Unitree’s quadrupeds and humanoids (Go2, B2, G1, H1), now common in legged robotics because they are cheap. (7 links)

Doppler velocity log: An acoustic sensor that measures an underwater vehicle’s velocity over the seabed. (12 links)
Lidar: Laser range sensing. (34 links)

COLMAP: Structure-from-motion and multi-view stereo software (Schönberger and Frahm, 2016). (10 links)
Gazebo: The open-source robot simulator most ROS projects use. (8 links)
Habitat: Meta’s simulator for embodied AI, built for navigation in scanned buildings such as HM3D. (9 links)
Kimera-Multi: Distributed multi-robot metric-semantic SLAM from MIT SPARK (Tian and others, 2022), the multi-robot version of Kimera. (8 links)
ORB-SLAM3: Feature-based visual and visual-inertial SLAM (Campos and others, 2021), the usual baseline for camera SLAM. (7 links)
ROS: The Robot Operating System: middleware for passing messages between a robot’s processes. (17 links)

CLIP: OpenAI’s contrastive image and text model (Radford and others, 2021). (19 links)
DINO: Meta’s self-supervised vision transformers (DINO, DINOv2, DINOv3). (12 links)
FastSAM: A fast segment-anything model built on a YOLO-style detector (Zhao and others, 2023), used where SAM is too slow for a robot. (8 links)
Gemini: Google DeepMind’s multimodal models, used as planners and failure judges in several of these papers. (9 links)
GPT-4: OpenAI’s large language and vision-language models. (17 links)

Gaussian splatting: A scene stored as millions of small 3D Gaussians that can be rendered in real time (Kerbl and others, 2023). (11 links)
Neural radiance fields: A scene stored as a neural network that maps a 3D point and a viewing direction to colour and density (Mildenhall and others, 2020). (6 links)
RANSAC: Random sample consensus (Fischler and Bolles, 1981): fit a model to random minimal subsets of the data and keep the one most points agree with. (12 links)

Linear temporal logic: A logic for properties of executions over time: “always”, “eventually”, “until”. (11 links)
PDDL: The Planning Domain Definition Language (McDermott and others, 1998), the standard input format for classical task planners. (8 links)

iTrust: SUTD’s centre for research in cyber security, home of the SWaT, WADI and EPIC testbeds, where I worked from 2018 to 2026. (8 links)
MARVL Lab: My lab at SUTD, led by Malika Meghjani. (3 links)
MIT SPARK Lab: Luca Carlone’s lab at MIT. (8 links)