Independent robotics intelligence
RoboSkin.ai tracks source-backed robotics research across robot skin, tactile sensors, robot hands, humanoid robots, dexterous manipulation, embodied AI, Physical AI, and visuo-tactile world models.
Tactile AI stack mapSurface / signal / inference / action — original RoboSkin.ai visual study
In practical robotics, robot skin helps robots detect contact, pressure, shear, slip, and interaction events across hands, grippers, arms, or curved body surfaces. For Physical AI, it is the contact layer that vision alone cannot provide.
Field map / Core authority
RoboSkin.ai maps the technologies, research, datasets, sensors, robot platforms, and AI models that power touch intelligence in robots. Start with a pillar, then follow its papers, datasets, benchmarks, and related entities.
Technologies, sensing principles, surface architectures, e-skin relationships, and research routes.
Map the sensing surface →The complete path from calibrated touch data to representations, models, robot control, and evidence.
Explore tactile intelligence →Full-hand, whole-arm, and whole-body tactile sensing for manipulation, interaction, and contact awareness.
Open the humanoid stack →Representation learning, tactile foundation models, visuo-tactile world models, policies, and transfer limits.
Compare emerging models →A filterable, source-reviewed database of tactile robotics data, sensors, robots, tasks, formats, and licenses.
Filter tactile datasets →How vision, language, proprioception, and touch combine in embodied perception, planning, and control.
Connect touch to Physical AI →AI / Robot relationship
Artificial intelligence supplies perception, prediction, reasoning, and action policies. Robotics supplies sensors, embodiment, controllers, actuators, and safety constraints. Their relationship becomes useful when physical outcomes return as feedback instead of ending at a generated command.
Open the AI and robotics field guide →Language defines the task while vision, robot state, and touch describe the current physical situation.
Open layer →Models turn images, joint state, force, and contact signals into representations the robot can use.
Open layer →A reasoning layer estimates state, plans task steps, and decides which capability should act next.
Open layer →An action model converts observations and instructions into trajectories, poses, or motor commands.
Open layer →Controllers and actuators turn a proposed action into timed movement on a specific robot body.
Open layer →Hands, grippers, feet, or body surfaces meet real objects, people, tools, and changing environments.
Open layer →Tactile AI detects contact, slip, force, and deformation so control or learning can respond to what actually happened.
Open layer →Physical contact returns evidence to perception, control, and learning. This is the layer where robot skin and tactile AI make a general robot model contact-aware.
Follow the learning loop →01 / Robotics research pulse
Broad robotics terms only earn useful authority when they connect to a clear evidence lane. RoboSkin.ai maps each large topic back to tactile sensing, robot hands, contact-rich tasks, data, models, and measurable limitations.
Research watch reviewed 2026-08-18
A new preprint equips a Unitree G1 with 60-element pressure insoles on each foot and feeds spatial and temporal load patterns into a locomotion policy. In the reported physical comparisons, Tac4Loco completed a ramp-to-foam transition in 10 of 10 trials versus 4 of 10 for the proprioception-only baseline.
The result expands humanoid robot skin beyond hands and arms: foot pressure verifies partial, asymmetric, or compliant support after touchdown, while vision remains the complementary pre-contact channel.
Track tactile coverage, dexterous hands, grasp stability, slip, and contact feedback for humanoid robot manipulation.
Explore humanoid robots →Map demonstrations, reinforcement learning, robot datasets, sim-to-real transfer, and tactile feedback into physical-world behavior.
Map robot learning →Compare visual, acoustic, magnetic, and resistive tactile sensing by contact-rich manipulation task and evidence boundary.
Map robot manipulation →Compare vision-language-action interfaces, embodiments, action outputs, real-robot evidence, artifact access, and tactile input.
Map robot VLA models →Latest source-backed updates
UniVTAC combines a tactile simulation platform, a 512-dimensional ResNet-18 representation encoder, and an eight-task benchmark—but its four data and evaluation pools must not be treated as one dataset.
A 2026 review maps vision-based tactile sensing as one integrated stack: deformable contact hardware, optical readout, tactile representations, learning, simulation, datasets, and robot action.
ADEPT reports 3/10 vision-only versus 8/10 visuo-tactile final success in one matched Flexiv-Sharpa insertion condition, with ten physical trials per condition.
02 / Signal to action
Research notes and resource entries organize the robot skin category around tactile sensors, e-skin architectures, stack maps, reader questions, and public reference paths.
Why tactile AI matters
Robots need contact data, not just vision, when tasks involve grasping, sliding, pressure, or safe physical interaction.
Read the application context →
Contact study / 2026Tactile AI stack map
Robot skin is useful when contact signals move through a complete stack: surface design, sensors, signal conditioning, robot middleware, controller behavior, safety response, and evaluation data.
Flexible, soft, stretchable, or conformal surfaces that define where contact can be measured.
Capacitive, piezoresistive, optical, magnetic, liquid metal, or multimodal sensor arrays for robot touch.
Filtering, calibration, timestamping, and feature extraction that turn raw contact into usable streams.
Local models and embedded processing for slip events, contact classification, and lower-latency response.
Middleware, controllers, and policies that use touch for grasping, safety, manipulation, and evaluation.
Contact-aware responses that help Physical AI systems behave more safely around people and objects.
Logs, datasets, benchmarks, and replay loops that make robot touch measurable and improvable over time.
03 / Research atlas
Use this research map to move from definitions to papers, technology evaluation, references, library pages, and source-submission paths.
In the RoboSkin context, Physical AI means physical-world AI systems that need robot skin, tactile AI, contact feedback, pressure, slip, and tactile sensing. The homepage is the broad research map; the Physical AI page is the canonical definition route.
Physical AI systems need local contact evidence when hands, grippers, tools, or body surfaces touch the world. Robot skin gives that evidence a surface layer.
Tactile AI connects pressure, shear, slip, calibration, timestamps, and controller-facing features so touch can support action, evaluation, or learning.
The strongest Physical AI route links visible definitions to tactile feedback, touch data, source-backed research, and conservative claim boundaries.
04 / Direct answers
Direct-answer coverage supports readers and answer engines without turning source boundaries into product claims.
Robot skin is a tactile sensing surface that helps robots detect contact, pressure, shear, slip, and interaction events across hands, grippers, arms, or curved body surfaces. It gives Physical AI systems a contact layer that vision alone cannot provide.
Open the robot skin glossary →Tactile AI is the sensing, data, and control workflow that turns touch signals into useful robot behavior. It can support grasp confidence, slip response, contact-aware motion, safety reflexes, and evaluation analytics for Physical AI systems.
Browse tactile AI research →Physical AI is a broad term for AI systems that perceive, reason, and act through physical machines. A complete system connects sensors and models to robot policies, control, actuation, safety, and measured feedback.
Read the Physical AI explainer →05 / Field guides
Use these public resources to navigate category research, stack maps, references, and source-backed learning paths.
RoboSkin.ai research asset
A reproducible record-level audit of data URLs, direct license links, code, sampling-rate, synchronization, and split disclosure.
RoboSkin.ai research asset
A versioned, source-reviewed index of robot-skin and tactile-AI work with explicit evidence classes and limitations.
RoboSkin.ai research asset
A hardware-neutral message contract, synthetic publisher, contract monitor, rosbag2 QoS, and calibration metadata example.
RoboSkin.ai research asset
Compare named task protocols, robots, sensors, metrics, access, and limitations without collapsing incompatible scores.
Start with the public research notes, suggest a source, or send a correction when a claim needs better support.
Submit research context →Primary path
Best for readers who want source-backed context on robot skin, tactile AI, and e-skin.
Partner path
Use for research collaboration, source suggestions, or educational content ideas.
Correction path
Use for unsupported claims, outdated references, or better public sources.
Direct inquiries: messigoat147@gmail.com