Robotic Exploration of Surfaces Using Tactile Data

Technology #14014

Questions about this technology? Ask a Technology Manager

Download Printable PDF

Image Gallery
Robotic exploration of unknown surfaces
Professor Steven Dubowsky
Department of Mechanical Engineering, MIT
External Link (
Francesco Mazzini
Department of Mechanical Engineering, MIT
Dan Kettler
Department of Mechanical Engineering, MIT
Julio Guerrero
Schlumberger Doll Research Center – Schlumberger Ltd.
Managed By
Christopher Noble
MIT Technology Licensing Officer - Clean and Renewable Energy
Patent Protection

Robotic exploration of unknown surfaces

US Patent Pending 2011-0029289
Tactile Robotic Mapping of Unknowning Surfaces, with Application to Oil Wells
IEEE Transaction on Instrumentation and Measurement, vol. 60, pp. 420-429.
An Experimental Validation of Robotic Tactile Mapping in Harsh Environment such as Deep Sea Oil Well Sites
12th International Symposium of Experimental Robotics, New Delhi, 2010
Tactile Robotic Mapping of Unknown Surfaces, with Application to Oil Wells
Robotic and Sensors Environments, ROSE 2009, IEEE International Workshop on 2009. 80-85


Mapping environments robotically is important for many applications including lowering instruments and tools into wells for oil exploration.

Problem Addressed

Previous technologies for mapping surfaces required complex force and torque sensors. This technology uses only relative positions of the robot's links. The procedure is faster than previous technologies because it minimizes the number of points it’s required to touch in order to define the geometry.


This technology has two parts, the robot and the algorithm the robot uses to map the surface. The robot uses a simple impedance controller to sense when it has contacted the surface. Then the position of the probe tip can be determined by sensing the robot's joint angles. The algorithm for the robot's exploration has two parts, surface reconstruction and exploration strategy. Based on the points already collected, the algorithm identifies the best approximate shape of the surface. For exploration, the robot starts in a random direction until the probe touches the surface and then probes the surface until a primitive is identified with desired accuracy. Then the robot chooses another random direction and moves until it touches a new point. If the second point is already part of a primitive then the robot picks another random direction and if not it will probe the surface and create a new primitive.


  • Does not require complex force or torque sensors
  • Increases mapping speed