Force-differential steering of a track-based robotic colonoscope: experimental evaluation in single and multi-bend pipes Authors: Jiayang Du, Lin Cao, Sanja Dogramazi Published: 2026 ABSTRACT This work presents a laboratory-scale track-based robotic colonoscope incorporating a force-differential steering mechanism. A syringe-driven hydraulic system selectively retracts individual tracks to generate asymmetric track-wall contact forces for steering. A theoretical model was developed to describe the relationships between hydraulic input, normal-force modulation, and steering torque and to provide a first-order framework for the mechanism. The prototype, was tested with force sensing and an IMU, was evaluated for straight propulsion and steering in single- and multi-bend flexible pipes. Hydraulic actuation tests showed a near-linear increase in track-wall normal force with increasing syringe input, although the measured force was substantially lower than the theoretical prediction. Steering experiments showed that motion without active steering produced approximately 35° of yaw before traction loss, whereas force-differential steering increased the yaw angle to approximately 45° and enabled negotiation of a 120° bend. Tests under different friction conditions further showed that reduced wall friction impaired steering performance. In a continuous S-shaped pipe, two of three independent trials successfully completed both bends, while one trial exhibited incomplete yaw recovery, indicating trial-to-trial variability and sensitivity to traction conditions. Overall, these results provide proof-of-concept evidence for force-differential steering in a tracked robotic colonoscope and demonstrate that straight propulsion and active planar steering can be achieved within a common tracked locomotion architecture. Further miniaturisation, improved hydraulic control, and evaluation in more physiologically representative environments are required before clinical applicability can be assessed. Introduction Stable navigation in the highly curved, compliant, and slippery colon cannot rely on propulsion alone, as insufficient steering control causes deviation from the centreline, unstable wall contact, and failure to reorient in bends (Santucci and Velez, 2024), which increases mucosal stress and prolongs the procedure (Eickhoff et al., 2010) (Alazmani et al., 2016). Active steering therefore plays a crucial role, and prior work has explored multiple strategies, including differential, articulated, magnetic assisted, and compliant continuum based approaches. Differential steering offers simplicity and fast response in non-medical pipe locomotion environments, but can be less efficient in high friction or small radius pipes (Roh and Choi, 2005; Oya et al., 2005; Elankavi et al., 2022). In the context of endoscopy, articulated steering can provide accurate orientation but is also limited by structural complexity and size constraints (Vucelic et al., 2006). Magnetic assisted steering enables wireless control but depends on large external magnets and suffers rapid force decay with distance (Wang et al., 2008) (Verra et al., 2020). Compliant or continuum steering achieves smooth bending through pressurized chambers, though slow response and deformation coupled actuation reduce its precision (Dehghani et al., 2017) (Chen et al., 2022) (Giri et al., 2025). Roh et al. (Roh and Choi, 2005) presented a differential drive in pipe robot controlled by multiple motors, yet its structurally complex multi-actuator configuration resulted in a bulky form difficult to scale down and therefore unsuitable for the highly confined and tortuous geometry of the colon. Dehghani et al. (Dehghani et al., 2017) developed a pneumatically driven self-steering colonoscope that could move at an average speed of 28 mm/s and achieve a minimum turning angle of approximately 30°. However, the steering performance relied on continuous pneumatic actuation and a relatively complex soft structure, which may pose challenges for miniaturization and precise control in confined colonic environments. Chen et al. (Chen et al., 2022) designed an inchworm inspired soft robotic colonoscope using rubber bellows, offering high compliance and safety but limited steering accuracy due to pneumatic hysteresis and structural constraints. Vucelic et al. (Vucelic et al., 2006) introduced the Aer-O-Scope system, which enabled full colon examination but retained an operation mode similar to conventional colonoscopy with limited autonomous navigation. As such, while the Aer-O-Scope demonstrates clinical feasibility, its steering capability remains largely operator-dependent, highlighting the need for robotic systems that can provide intrinsic and autonomous directional control. Giri et al. (Giri et al., 2025) developed the InchIGRAB mechanism combining guided traction with continuum bending, capable of operating through bends of 80°–130°, though its multi-chamber design increased synchronization complexity and reduced response speed. Recent advances in compliant continuum robots and snake-like robots have further improved adaptability and manoeuvrability for navigation in confined environments through compliant structures and hyper-redundant locomotion strategies (Liu et al., 2021; Burgner-Kahrs et al., 2015). Although these systems have achieved notable progress, the variable curvature and diameter of the colon, its compliant low-friction wall, and strict limits on allowable contact pressure mean that current approaches remain inadequate in terms of size, control precision, response speed, and the ability to simultaneously achieve reliable propulsion and steering. Consequently, they do not fully satisfy the combined mechanical and control requirements imposed by the colonic environment. In our earlier work (Du et al., 2025), we demonstrated that a track-based robotic colonoscope with a bellows-driven expansion mechanism could effectively modulate track-wall contact pressure to maintain continuous contact over varying colon diameters and provide the basis for generating steering torque. Building on this foundation, the present study replaces the original bellows-based expansion mechanism with a syringe-driven hydraulic actuation system, enabling controllable track retraction and asymmetric wall contact for force-differential steering. Compared with representative robotic colonoscope steering strategies, the proposed system provides a favourable balance between steering capability, propulsion performance, mechanical simplicity, and size adaptability. The laboratory-scale prototype successfully negotiates bends of up to 120° while maintaining continuous tracked propulsion, achieves a stable yaw angle of approximately 45° during active steering, and operates over an experimental pipe-diameter range of 95–160 mm. Rather than maximising locomotion speed, the proposed design prioritises steering capability and geometric adaptability, which are critical requirements for negotiating tortuous colonic pathways. Compared with the InchIGRAB mechanism (Giri et al., 2025), which negotiates bends of 80°–130° using a 116 mm front section incorporating a 65 mm pneumatic actuation module with a diameter of 26 mm, the proposed robot achieves comparable bend negotiation through a simpler force-differential steering mechanism integrated into a compact tracked platform without differential track speeds, articulated joints, or external magnetic actuation. This relatively simple architecture enables controllable wall-pressure redistribution while reducing mechanical complexity and maintaining good potential for future miniaturisation. In this mechanism, selective hydraulic retraction of individual tracks creates an asymmetric distribution of track-wall contact forces, thereby generating a yaw moment while maintaining continuous propulsion. Steering performance is evaluated primarily in terms of planar yaw motion to investigate the feasibility and characteristics of force-differential steering in curved pipes. The following sections describe the operating principle, mathematical modelling, and experimental validation of the proposed approach, and evaluate its steering performance under different geometric and frictional conditions. System design and steering mechanism Compared with the first prototype, the second prototype introduces several improvements in structural design, actuation mechanism, and overall system reliability (Du et al., 2025; Du et al., 2024; Evans et al., 2025). First, the pneumatic bellows based expansion mechanism used in (Du et al., 2025) was replaced with a syringe based hydraulic actuation system. Compared with pneumatic actuation, the hydraulic system uses incompressible liquid for pressure transmission, providing more stable output forces, faster response characteristics, and reduced delay caused by air compressibility. In addition, the original silicone tracks were replaced with TPU (thermoplastic polyurethanes) tracks, improving structural rigidity, wear resistance, and frictional traction against the pipe wall, thereby enhancing propulsion performance and motion stability. Four tracks were selected as a practical compromise between structural simplicity, circumferential support, and steering capability. The symmetric 90° arrangement provides balanced contact around the lumen and allows force-differential steering through independent adjustment of the contact force of each track. Although alternative configurations with different numbers of tracks are possible, their comparative evaluation is beyond the scope of the present study. From the mechanical perspective, the second prototype further optimised both the transmission and supporting structures. The micro bearings were introduced at both ends of the flexible shaft to improve rotational stability and reduce friction losses during operation, thereby improving torque transmission efficiency and overall system stability as shown in Figure 1. Furthermore, to address the relatively high sliding friction observed between the tracks and the internal supporting structure in the first prototype, a dedicated low friction guiding support structure was designed and integrated into the new prototype system. This structure provides a more stable guiding interface between the inner side of the tracks and the robot body, reducing unnecessary friction and local jamming during track motion, which reduces energy losses and movement jerks. (A) The new robotic colonoscope prototype; (B) structural configuration of the robot head; (C) low friction guiding support structure for track stabilisation. (D) modular track support and mounting structure. The track constraint and mounting structures were also further refined. Optimised arc shaped guiding grooves were incorporated into the robot body to limit lateral displacement of the tracks and improve track stability during locomotion. Lubricated contact surface were additionally introduced to further minimise internal friction resistance. To improve maintenance efficiency and experimental repeatability, the second prototype adopts a modular quick release structure, allowing the track modules and expansion mechanisms to be more easily assembled, replaced, and maintained. The track width was also increased from 7.5 mm to 12 mm to further improve the contact area and traction force during locomotion. These structural and actuation related improvements collectively enhance the propulsion speed, traction and steering stability, enabling more reliable locomotion in complex curved tubular environments. Figure 1 shows the complete new robotic prototype system, which consists of the robot head, an external motor transmission system, a syringe based hydraulic actuation system, and an electronic control system. The robot head is connected to an external motor through a flexible shaft, allowing the motor torque to be transmitted to an internal worm mechanism that synchronously drives the four tracks. The hydraulic actuation system transfers hydraulic pressure from external syringes to miniature syringes inside the robot, enabling radial deformation of each track module for lumen adaptation and active steering based on differential contact forces. The entire system is coordinated by a microcontroller based control platform, which controls both the propulsion motor and the hydraulic actuation units. This externalised actuation architecture reduces the number and mass of the components that must be accommodated within the robot head, while simplifying internal wiring and improving system modularity and experimental accessibility. However, the current device is a proof-of-concept prototype developed to evaluate the force-differential steering principle and does not represent a clinically miniaturized colonoscope. Principle of steering mechanism The proposed robotic colonoscope uses four evenly distributed tracks driven synchronously by a single motor through a flexible shaft and worm gear, ensuring equal linear velocity during propulsion. This has been shown in our previous work (Du et al., 2025). Differential steering is achieved by regulating the normal forces between the tracks and the lumen wall through a syringe driven hydraulic expansion mechanism. Hydraulic input causes each track to expand or retract radially, and selectively retracting one side creates an asymmetric wall pressing distribution that generates a yaw moment. During straight motion the forces remain balanced, while during steering the reduced normal force on one side and the sustained traction on the other induce rotation until the robot realigns with the lumen, after which pressures on the four tracks are equalized. This force differential steering replaces speed differences with controlled variations in wall pressure, and the resulting yaw torque depends on both the magnitude of the normal force imbalance and the effective moment arm. By precisely regulating hydraulic input, the robot maintains stable wall contact and performs smooth, continuous, and reversible steering, enabling reliable navigation through narrow, slippery, and curved lumens. Mathematical modelling of the steering mechanism Geometric constraint analysis When traveling through a curved section, the robot must satisfy the spatial constraints imposed by the pipe curvature; if its width or length is too large relative to the lumen diameter and curvature radius, it may fail to steer around the bend. A geometric constraint analysis is therefore required to identify the admissible ranges of the width and length that allows smooth passage through curved segments without collision or excessive wall deformation. Figure 2 illustrates the geometric relationship between the robot and the curved pipe, where R denotes the pipe curvature radius, D is the local lumen diameter at the considered cross-section. For simplicity, the local diameter is assumed to remain constant within the geometric analysis, although it varies spatially in the human colon, as shown in Equations 1, 2. W is the robot width, L is the robot length, and γ represents the bending angle of the curved section. Based on this configuration, the following geometric constraints can be derived to describe the feasible range of W and L . Geometric model approximation of the robot in a curved section. The width, W, of the robot has to satisfy the following relationship: R+D2cos⁡γ−R−D2