DogTouch: CNN-based Recognition of Surface Textures by Quadruped Robot with High Density Tactile Sensors

The ability to perform locomotion in various terrains is critical for legged robots. However, the robot has to have a better understanding of the surface it is walking on to perform robust locomotion on different terrains. Animals and humans are able to recognize the surface with the help of the tactile sensation on their feet. Although, the foot tactile sensation for legged robots has not been much explored. This paper presents research on a novel quadruped robot DogTouch with tactile sensing feet (TSF). TSF allows the recognition of different surface textures utilizing a tactile sensor and a convolutional neural network (CNN). The experimental results show a sufficient validation accuracy of 74.37% for our trained CNN-based model, with the highest recognition for line patterns of 90%. In the future, we plan to improve the prediction model by presenting surface samples with the various depths of patterns and applying advanced Deep Learning and Shallow learning models for surface recognition. Additionally, we propose a novel approach to navigation of quadruped and legged robots. We can arrange the tactile paving textured surface (similar that used for blind or visually impaired people). Thus, DogTouch will be capable of locomotion in unknown environment by just recognizing the specific tactile patterns which will indicate the straight path, left or right turn, pedestrian crossing, road, and etc. That will allow robust navigation regardless of lighting condition. Future quadruped robots equipped with visual and tactile perception system will be able to safely and intelligently navigate and interact in the unstructured indoor and outdoor environment.


page 1

page 3

page 4


Walking on TacTip toes: A tactile sensing foot for walking robots

Little research into tactile feet has been done for walking robots despi...

Surface Following using Deep Reinforcement Learning and a GelSightTactile Sensor

Tactile sensors can provide detailed contact in-formation that can facil...

MobileCharger: an Autonomous Mobile Robot with Inverted Delta Actuator for Robust and Safe Robot Charging

MobileCharger is a novel mobile charging robot with an Inverted Delta ac...

DeltaCharger: Charging Robot with Inverted Delta Mechanism and CNN-driven High Fidelity Tactile Perception for Precise 3D Positioning

DeltaCharger is a novel charging robot with an Inverted Delta structure ...

Can Quadruped Navigation Robots be Used as Guide Dogs?

Bionic robots are generally considered to have strong flexibility, adapt...

Wait, That Feels Familiar: Learning to Extrapolate Human Preferences for Preference Aligned Path Planning

Autonomous mobility tasks such as lastmile delivery require reasoning ab...

An RFID-Based Assistive Glove to Help the Visually Impaired

Recent studies have focused on facilitating perception and outdoor navig...

Please sign up or login with your details

Forgot password? Click here to reset