In the field of robotics, achieving robust mobility across unstructured terrains remains a significant challenge. Traditional mobile robots, such as wheeled or tracked platforms, often struggle with complex obstacles like trenches and steps, while legged robots, though adaptable, face issues in control complexity and speed. To address these limitations, we draw inspiration from nature, specifically from the turtle, an amphibious reptile known for its stable crawling and ability to traverse rough landscapes. In this work, we present the design, gait planning, and obstacle-surmounting performance analysis of a novel wheel-track-leg composite bionic robot. This bionic robot integrates the advantages of wheels, tracks, and legs to enhance adaptability and stability in diverse environments.
The core innovation lies in the hybrid mobility mechanism, which allows the bionic robot to switch between different locomotion modes based on terrain demands. We begin by detailing the mechanical design, inspired by the turtle’s body structure and stability mechanism. The bionic robot features a main body equipped with four independently actuated legs, each comprising a thigh, a shank, and a swamp wheel assembly. Each leg has two degrees of freedom: horizontal rotation at the hip joint (driven by a swing motor) and vertical rotation at the knee joint (driven by a leg motor). This configuration enables the bionic robot to mimic the turtle’s crawling motions. Additionally, the bionic robot incorporates a track system for efficient movement on flat surfaces and a rotary support for attaching manipulators or other payloads. The following table summarizes the key structural parameters of the bionic robot:
| Parameter | Value |
|---|---|
| Mass (kg) | 9.8 |
| Dimensions (L × W × H, mm) | 527 × 376 × 135 |
| Thigh Length (mm) | 51 |
| Shank Length (mm) | 173 |
| Swamp Wheel Radius (mm) | 37.5 |
| Track Load Wheel Spacing (mm) | 86 |
| Track Load Wheel Radius (mm) | 24 |
To control the bionic robot effectively, we establish a kinematic model using the Denavit-Hartenberg (D-H) method. For each leg, we define coordinate frames attached to the joints and end-effector (the swamp wheel center). Taking the right-front leg as an example, the base frame {O} is fixed at the robot’s center of gravity. The transformation matrix between consecutive frames for leg i (i=1,2,3,4) is derived. For joint j (j=1,2) of leg i, the D-H parameters include link length lij, twist angle εij, offset dij, and joint angle θij. The homogeneous transformation matrix Aij is given by:
$$A_{ij} = \begin{bmatrix} \cos\theta_{ij} & -\cos\varepsilon_{ij}\sin\theta_{ij} & \sin\varepsilon_{ij}\sin\theta_{ij} & l_{ij}\cos\theta_{ij} \\ \sin\theta_{ij} & \cos\varepsilon_{ij}\cos\theta_{ij} & -\sin\varepsilon_{ij}\cos\theta_{ij} & l_{ij}\sin\theta_{ij} \\ 0 & \sin\varepsilon_{ij} & \cos\varepsilon_{ij} & d_{ij} \\ 0 & 0 & 0 & 1 \end{bmatrix}$$
The position of the foot tip Bi in the base frame is obtained by cascading transformations. For leg 1, the forward kinematics yields:
$$\begin{cases} B_{1x} = l_1 \cos \theta_{11} + l_2 \cos \theta_{11} \cos \theta_{12} + u \\ B_{1y} = l_1 \sin \theta_{11} + l_2 \sin \theta_{11} \cos \theta_{12} + v \\ B_{1z} = l_2 \sin \theta_{12} – w \end{cases}$$
where (u, v, -w) is the origin of the leg’s initial frame in {O}, and l1, l2 are thigh and shank lengths, respectively. Similar equations are derived for other legs, enabling comprehensive motion planning for this bionic robot.
Gait planning is crucial for the bionic robot to emulate turtle-like crawling. By observing turtle locomotion, we decompose a single gait cycle into four phases: leg release, crawling, leg lift, and leg swing. Based on this, we plan four distinct gaits for the bionic robot, as summarized in the table below:
| Gait Type | Description | Leg Involvement | Primary Motion |
|---|---|---|---|
| Two-Leg Crawling Gait | Only front or rear legs actuate while opposite legs provide support. | 2 legs | Forward/Backward Movement |
| Four-Leg Crawling Gait | All four legs actuate synchronously, maintaining symmetry. | 4 legs | Forward/Backward Movement |
| Rotational Gait | Legs on one side move forward while opposite side moves backward. | 4 legs | In-Place Rotation |
| Lateral Gait | Right or left-side legs actuate while others support. | 2 legs | Lateral Movement |
In the four-leg crawling gait, for instance, the bionic robot starts with the body on the ground and wheels lifted. The legs simultaneously lower to contact the ground, lift the body, then swing backward to propel the bionic robot forward via static friction from the wheels. After reaching a new position, the legs lift, the body lowers, and the legs swing forward to reset. This cyclic process allows continuous crawling. The joint torques required during static equilibrium in this gait are analyzed. Let Ni and fi be the ground reaction force and static friction at foot Bi, respectively. For symmetry, θf1 = θ11 = θ21, θr1 = θ31 = θ41, and θf2 = θr2 = θi2. The torque at joint 1 (Ti1) and joint 2 (Ti2) for leg i are:
$$T_{i1} = f_i (l_1 + l_2 \cos \theta_{i2}), \quad T_{i2} = N_i l_2 \cos \theta_{i2}$$
where fi = μNi, with μ as the static friction coefficient. This analysis aids in motor selection for the bionic robot.
The obstacle-surmounting capability is a key metric for this bionic robot. We evaluate performance by deriving theoretical models for crossing trenches and climbing steps, focusing on maximum achievable widths and heights. For trench crossing, the bionic robot must maintain stability to prevent tipping. The process involves three stages: entry, traversal, and exit. Assuming symmetric leg configurations (θi1 = 90°, θi2 = -23° during crossing), we define conditions for each stage. Let L1, L2, L3 be the maximum trench widths satisfying stability in stages 1, 2, and 3, respectively:
$$\begin{cases} L_1 = B_{fy} \\ L_2 = 5d \\ L_3 = -B_{ry} \end{cases}$$
where Bfy and Bry are the Y-coordinates of front and rear foot tips in {O}, and d is the track load wheel spacing. The maximum crossable trench width Lw is:
$$L_w = \min(L_1, L_2, L_3)$$
For step climbing, the bionic robot uses a combination of leg lifting and track propulsion. The process is divided into three phases: initial contact, mid-climb, and final ascent. In phase 1, the front legs lift to ensure the swamp wheels clear the step height H1:
$$H_1 = h + B_{fz} – R$$
where h is the vertical distance from the center of gravity to the track base, Bfz is the Z-coordinate of the front foot, and R is the swamp wheel radius. In phase 2, the front wheels engage the step, and the robot pivots around the rear track wheels. The angle α between the line connecting rear wheel center Ar to the base frame origin O and the horizontal must be less than 90° to avoid tipping. It is expressed as:
$$\alpha = \alpha_1 + \alpha_2 + \alpha_3$$
with:
$$\alpha_1 = \arcsin\left(\frac{h – r}{l_{AO}}\right), \quad \alpha_2 = \arctan\left(\frac{r – h – B_{fz}}{B_{fy} + l_{AO} \cos \alpha_1}\right), \quad \alpha_3 = \arcsin\left(\frac{H_2 + R – r}{l_{AB}}\right)$$
where r is the track wheel radius, lAO and lAB are distances, and H2 is the maximum step height for this phase. In phase 3, the rear legs actuate to push the robot onto the step. The condition for successful ascent is that the projection of the center of gravity remains within the step surface when aligned vertically with the step edge. The maximum step height H3 depends on leg angles θf2 and θr2:
$$H_3 = \omega(\theta_{f2}, \theta_{r2})$$
where ω is a function derived from geometric constraints. The overall maximum climbable step height H is:
$$H = \min(H_1, H_2, H_3)$$
To validate our design and models, we fabricated a prototype of the bionic robot and conducted experiments on granite and concrete surfaces. The bionic robot successfully executed all planned gaits, demonstrating smooth transitions in lateral, longitudinal, and rotational movements. For obstacle negotiation, the bionic robot was tested on trenches and steps. The experimental results for trench crossing showed a maximum width of 434 mm, slightly exceeding the theoretical value of 430 mm (relative error 0.92%), confirming the trench model’s accuracy. For step climbing, multiple trials with different leg configurations were performed, as summarized below:
| Trial | Front Leg Angle θf2 (°) | Rear Leg Angle θr2 (°) | Theoretical Max Height H (mm) | Experimental Height H* (mm) | Relative Error (%) |
|---|---|---|---|---|---|
| 1 | -30 | -30 | 36.1 | 34 | 6.17 |
| 2 | -45 | -45 | 103.1 | 98 | 5.20 |
| 3 | -60 | -60 | 150.0 | 144 | 4.17 |
| 4 | -90 | -90 | 178.3 | 175 | 1.88 |
| 5 | -45 | -60 | 126.2 | 123 | 2.60 |
| 6 | -45 | -90 | 139.0 | 137 | 1.46 |
| 7 | -60 | -45 | 129.7 | 124 | 4.60 |
| 8 | -90 | -45 | 164.4 | 158 | 4.05 |
The bionic robot achieved a maximum step height of 175 mm when both front and rear legs were depressed at -90°, closely matching theoretical predictions. Errors are attributed to dynamic effects like slight body oscillations and minor shifts in the center of gravity during leg actuation. Overall, the experiments verify the feasibility of the bionic robot’s mechanical design, the effectiveness of the turtle-inspired gaits, and the correctness of the obstacle-surmounting models.
In discussion, we emphasize that this wheel-track-leg composite bionic robot offers a balanced solution for terrain adaptability. The integration of multiple mobility mechanisms allows the bionic robot to leverage wheels for speed on flat ground, tracks for traction on soft surfaces, and legs for overcoming discrete obstacles. The bionic robot’s stability, derived from the turtle-like wide stance, ensures reliable operation during complex maneuvers. Furthermore, the gait planning framework provides flexibility for autonomous navigation in unstructured environments. Compared to conventional robots, this bionic robot demonstrates superior obstacle negotiation without compromising control simplicity, as the gaits are based on synchronized joint motions rather than complex dynamic algorithms.
Looking forward, we envision several enhancements for this bionic robot. Incorporating sensors like LiDAR or cameras could enable real-time terrain perception and adaptive gait selection. Advanced control strategies, such as model predictive control, might improve dynamic performance during high-speed operations. Additionally, material optimization could reduce weight, increasing energy efficiency. The modular design of the bionic robot also allows for integration with robotic arms or other tools, expanding its applications in search-and-rescue, exploration, or industrial inspection.
In conclusion, we have presented a comprehensive study on a wheel-track-leg composite bionic robot, inspired by turtle morphology and locomotion. The bionic robot’s design combines wheels, tracks, and legs to achieve versatile mobility. Through kinematic modeling, we derived motion equations for precise control. Four turtle-inspired gaits were planned, enabling the bionic robot to perform lateral, longitudinal, and rotational movements. Theoretical models for trench crossing and step climbing were developed, predicting maximum obstacle dimensions that were validated through experiments. The bionic robot prototype demonstrated successful gait execution and obstacle negotiation, with a maximum trench width of 434 mm and step height of 175 mm. This work confirms the viability of hybrid bionic robots for challenging terrains and provides a foundation for future advancements in adaptive robotic systems. The bionic robot represents a significant step toward autonomous machines capable of thriving in diverse real-world conditions.