Autonomous control for miniaturized mobile robots in unknown pipe networks

Abstract

Despite recent advances in robotic technology, sewer pipe inspection is still limited to conventional approaches that use cable-tethered robots. Such commercially available tethered robots lack autonomy, and their operation must be manually controlled via their tethered cables. Consequently, they can only travel to a certain distance in pipe, cannot access small-diameter pipes, and their deployment incurs high costs for highly skilled operators. In this paper, we introduce a miniaturised mobile robot for pipe inspection. We present an autonomous control strategy for this robot that is effective, stable, and requires only low-computational resources. The robots used here can access pipes as small as 75 mm in diameter. Due to their small size, low carrying capacity, and limited battery supply, our robots can only carry simple sensors, a small processor, and miniature wheel-legs for locomotion. Yet, our control method is able to compensate for these limitations. We demonstrate fully autonomous robot mobility in a sewer pipe network, without any visual aid or power-hungry image processing. The control algorithm allows the robot to correctly recognise each local network configuration, and to make appropriate decisions accordingly. The control strategy was tested using the physical micro robot in a laboratory pipe network. In both simulation and experiment, the robot autonomously and exhaustively explored an unknown pipe network without missing any pipe section while avoiding obstacles. This is a significant advance towards fully autonomous inspection robot systems for sewer pipe networks.

Keywords: autonomous control; exhaustive search; exploration; in-pipe robot; infrastructure robot; miniature robot; navigation; water & sewer pipes.

FIGURE 1

Robot CAD model and current physical model used in this paper. The main difference between the two models is that the physical robot does not need/equip an ESP32-CAM (development board for streaming camera video via WiFi) and has a simplified upper shell (for convenient downloading of the program for experiments).

FIGURE 2

Diagrams of Joey robot operating inside a schematic of our experimental pipe network. Panel (A) depicts the Joey robot at starting point (Entrance E) of a typical imitation in laboratory setting of a real sewer pipe network at the Integrated Civil and Infrastructure Research Centre, University of Sheffield (ICAIR). The imitated pipe network includes features of the real network such as T-junction, corners, branches, dead-end and obstacle. Panel (B) illustrates the robot coming from the left side of the network in Panel (A). Panel (B) also illustrates the left/right ranging direction which form angles of 30° with robot front-back axis.

FIGURE 3

Possible robot states in sewer pipe networks. Each arrow represents a possible state transfer. Junction-like states (i.e., branches, cross, corners, dead end) do not transfer directly to each other because of their distance from each other in real sewer pipe networks. They often interchange to straight pipe states, or sided states during autonomous navigation. These three states are interchangeable but have similar conditions thus are grouped together. Another group includes the risk states (i.e., crash risks, and flip risk). The risk states often occur during maneuvering at junctions.

FIGURE 4

Principles of estimating robot state by range sensors. Panel (A) shows the relative position of the Joey robot base in the round pipe radius R. r₁ is the half pipe width at the height that range sensor measuring ray points to. Panel (B) and Panel (C) depict the two typical orientations of the Joey approaching a right branch and the principles of determining a right branch with range sensors. In both cases, the right range value is significantly larger than the expected right range value for a straight pipe.

FIGURE 5

Junction classification using range sensors. Panel (A) and Panel (B) show the robot in a straight pipe, with similar left and right range values (these two states can be distinguished by checking the robot inertia data). Panel (C) illustrates a typical position of the robot when it detects a dead end (its front range value is under a certain limit and the sum value of its three range sensors is under a defined limit). Panel (D) shows the robot approaching a corner. As its left range value is significantly higher than the expected value of the straight pipe case, it can determine there is a left corner. Panel (E) and Panel (F) show the robot approaching a T-junction at a straight angle (α = 0, Panel E) or at a small angle (α > 0, Panel F). By changing this small angle α, we can determine the minimum and maximum distance ${\mathit{d}}_{\mathit{A}}$ (i.e., T_ASSESSING_DIS_MIN and T_ASSESSING_DIS_MAX) to assess if the robot is approaching a T-junction or corner.

FIGURE 6

Laboratory version of small diameter (150 mm) sewer pipe network. The pipe network was constructed with a T-junction (or left branch, right branch depending on the robot moving direction as showed by the arrows), a left/right corner, a dead-end, an obstacle and three straight pipe sections. One straight section was cut half open for visual observation.

FIGURE 7

Decisions and maneuvers of the Joey robot at right and left branches during autonomous exhaustive exploration. The top two rows show in sequence (RB-1 to RB-8) images of Joey approaching and then turning into a right branch. The bottom row shows in sequence (LB-1 to LB-4) the captured images of Joey approaching but not turning into a left branch. The robot continues straight at left branch, in agreement with the high-level rule of taking the rightmost direction.

FIGURE 8

Successful detection, decision making and maneuvers of the Joey robot at right (RC-1 to RC-8, top two rows) and left (LC-1 to LC-8, bottom two rows) corners during autonomous pipe exploration.

FIGURE 9

Successful detection, decision making and maneuvers of the Joey robot at T-junction (T-1 to T-8, top two rows) and large obstacle (OB-1 to OB-8, bottom two rows) in autonomous pipe exploration.

FIGURE 10

Snapshots of the Joey robot locomoting across different surfaces and on an inclined pipe. TER-1 shows the robot moving across dry sand. TER-2 and TER-3 show the robot moving along pipes with dish washing liquid, and a mixture of this liquid and sand. Lower panels: Joey robot climbing up (INC-4) and down (INC-3) an inclined pipe section shown in INC-1 and INC-2.