Optical Tactile Sensing for Aerial Multi-Contact Interaction: Design, Integration, and Evaluation

TL;DR

An optical tactile sensor that features a large and curved soft-sensing surface, a hollow structure and a new illumination system that provides real-world quantities of 3-D contact locations and 3-D force vectors is proposed.

Abstract

Distributed tactile sensing for multi-force detection is crucial for various aerial robot interaction tasks. However, current contact sensing solutions on drones only exploit single end-effector sensors and cannot provide distributed multi-contact sensing. Designed to be easily mounted at the bottom of a drone, we propose an optical tactile sensor that features a large and curved soft sensing surface, a hollow structure and a new illumination system. Even when spaced only 2 cm apart, multiple contacts can be detected simultaneously using our software pipeline, which provides real-world quantities of 3D contact locations (mm) and 3D force vectors (N), with an accuracy of 1.5 mm and 0.17 N respectively. We demonstrate the sensor's applicability and reliability onboard and in real-time with two demos related to i) the estimation of the compliance of different perches and subsequent re-alignment and landing on the stiffer one, and ii) the mapping of sparse obstacles. The implementation of our distributed tactile sensor represents a significant step towards attaining the full potential of drones as versatile robots capable of interacting with and navigating within complex environments.

Figures (9)

• Figure 1: Optical tactile sensing for drones. A quadrotor embodying the proposed optical tactile sensor to detect multiple contacts and interaction forces during physical interaction with the environment.
• Figure 2: Mechanical design and main components of the optical tactile sensor. (A) Developed prototype of optical tactile sensor shaped as an arc of a circular ring, integrated beneath a quadrotor (side and front view). The sensing area (green) is on the outer side of the sensor. The sensor coordinate frame is highlighted. (B) Assembled sensor with internal view exposed. (C) Fully assembled sensor with paper shield on the sides; sensing area marked in different contact locations. (D) Exploded view of all the components: 1) Sensing area (acrylic substrate and silicone material) with two connectors glued on; 2) lateral panels and 3) front panels for narrow side with M2 brackets and the other connectors screwed on; 4) UV LEDs strips; 5) USB camera; 6) color filter with holder; 7) camera holder and sensor base; 8) connector to attach the sensor to the drone.
• Figure 3: Flow-diagram for force and contact estimation. (A) Software pipeline developed for our optical tactile sensor: libraries shown in blue, mathematical functions in gray, contribution in green. (B) Graphical user interface: on the left, raw and divergent optical flow, potential field, and cropped camera image are displayed; on the right, the parameters used for cropping and thresholding can be changed to see the effect on the sensor performances.
• Figure 4: The setup used for characterization of the optical tactile sensor. (A) The sensor is attached to a base plate, placed on the table; the base can be tilted to rotate the sensor in multiple orientations. A load cell is connected to the vertical axis of a 3D printer, at the top, and to the indenter, at the bottom. A custom software framework allows to autonomously press the indenters on the sensing area. (B) Closeup of the cylindrical indenter with diameter equal to 10 mm pressing on the sensor. The indenter covers the whole sensor width since it is centered with respect to it. (C) Indenters used for the characterization; diameter of the cylinders is equal to 10 mm, 5 mm, and 2mm. (D) 2-axis tilting base used to hold the sensor in different orientations. Pitch and yaw rotations are reported. (E) The acquisitions are performed on the sensing area along the x-axis, in the center of the sensor's width, i.e. measurements on the numbered rows and centered on the dashed line.
• Figure 5: Estimation of a single contact continuously applied at position 5. Ground truth from the load cell, force and contact location estimated (and filtered) online with our software pipeline.