It was truly a pleasure to work with Axel as the project grew to draw from the diverse talents of the JPL community. The following people, members of the Axel team past and present, deserve much of the credit for the rover's recent and remarkable achievements: Jaret Matthews, Robert Peters, Jack Morrison, Benjamin Solish, Bob Anderson, Jack Dunkle, Srikanth Saripalli, Pan Conrad and Robert Miyake.
A Brief History of Space Robotics
Ranger Program In 1961, unmanned impact vehicles Rangers 1-6 attempted to take close-up photographic images of the moon before impacting the surface. Several Surveyor missions also carried robotic shovels designed to test the stability of the Moon's regolith1.
Future Science Targets
Exploring these areas and accessing the source of the methane plumes will be an important objective for future Mars rovers. Each of the aforementioned locations would serve as an excellent target for future robotic missions.
The Challenge of Robotic Operation in Extreme Terrain
A Survey of Extreme Terrain Rovers
Despite this ability to overcome high slip on slopes, steeper slopes are likely to require an external force, such as that generated by the use of a chain. Lateral mobility with two chains will generally be greater at a short distance from the rim, but this advantage diminishes as the rover descends deeper into the crater.
Thesis Outline and Contributions
The analysis is used to provide justification for a uniaxial tethered rover concept to provide access to science targets in extreme terrain areas. Following the single-axis concept, the development of the first-generation Axel prototype is described, along with field experiments that test and validate the rover.
Mission Architecture Trade-Offs
- Mother-Daughter System
- Single-Vehicle System
- Multiple Mobile Assets
- Multiple Fixed Assets
- Architecture Selection
Although the second option allows for a slightly simpler design, the mission would require a precision landing with a landing uncertainty ellipse in the hundreds of meters, and the daughter ship would be limited to surveying the area around the mother ship, limited by the length of the ship. berth. The first option, where a mother would carry her daughter to the edge of extreme terrain, would only require a precise landing with a multi-kilometer ellipse of uncertainty.
The Axel Concept
By passing the tire through the wheel arm and turning it around a central cylinder, Axel's own body acts as the pulley of a winch. Unwinding in this manner, the rope is laid across the terrain as Axel descends, and it is picked up as the rover returns to the host.
First-Generation Axel Prototype
Proof of Concept
In roll mode, the Axel wheels are powered and the wheel arm is held fixed, keeping the pitch angle of the body constant as the rover moves. In glide mode, the roll arm is activated while the wheels are held fixed, causing the body to fall forward as it moves.
Artificial Lunar Crater Tests
The new sampling mechanism used a design similar to the previously used T-shaped device, in which the sampling tube openings remained perpendicular to the long axis of the swing arm. The comb wheels were very effective at overcoming obstacles on the Mars Yard, easily traversing rocks up to 55% of the wheel diameter, with and without the aid of the chain (Figure a) Axel 1 on a vertical plane showing the main features with the label (b) Axel 1 Taking a soil sample.
Cleated Wheels
Although bicycle tires had metal studs, they had a much lower profile than aluminum rims.
Grouser Wheels
The comb wheels not only ensure the stability of the rover on steep slopes, but also provide excellent ability to overcome obstacles. The cam wheels developed for Axel performed well in the Mars Yard, but these important wheel parameters must be adjusted based on the rover's specific mission requirements.
Analysis and Experiments
Obstacle Traversal
Note that as θπ/2 approaches, the magnitude of µ must tend toward infinity if the wheel is to climb over the obstacle. To allow the wheel to pass over the obstacle without slipping, note that α = −ay/(r+l) and again|f|<|µn|. Note that since the lower limit of the friction coefficient is inversely proportional to the input torque, this limit can be decreased by increasing the wheel torque.
Sinkage
You can calculate the force on each support as a function of a single variable, the distance from the bottom of the rim to the ground. In static equilibrium, the force from the struts will balance with the weight of the rover and that. a) A wheel with two grooves sunk into the ground. The data are plotted against the theoretical model in Figure 3.7, where the wheel height is calculated as a function of the large groove angle from the vertical, β.
Efficiency
Note the large amplitude and sinusoidal nature of the power curve for grouser wheels. This corresponds to the increased torque required by the motor to push the wheel to the edge of the pit and then roll forward once the weft has passed vertical. The difference in linear speed between the two wheels is a result of the grouser wheels having a slightly larger effective radius.
Paddle-Rim Wheels
The Wheeled Wheel introduced an adaptive wheel design that, if activated, could alternate between modes for efficient long-distance travel and obstacle traversal. Although radio signals can reach the Moon in about two seconds, communication with rovers on other planets varies greatly due to the elliptical orbits of the celestial bodies. The chapter concludes with a simple example and a full simulation of the algorithm on terrain data of the Shackleton Crater of the Moon acquired by the Lunar Reconnaissance Orbiter.
Related Work
To navigate the Martian surface, robotic rovers need sophisticated motion planning algorithms that incorporate sensor data to first detect obstacles and then plan safe paths around them. The motion planning problem addressed in this thesis is unique in that not only must entanglement of the tether be avoided, but the tether and wheels working together must also be able to generate sufficient forces to ascend or descend steep slopes. While much of the design framework developed in this chapter could be applied to anchor winch systems, the focus is on robot-side winch systems.
Rover Dynamics
Holonomic and Non-Holonomic Constraints
Holonomic constraints limit the possible configurations of a system to a subvariety of the configuration space, Q. As mentioned earlier, constraints are enforced by forcing forces that affect the motion of the system. M(q)¨q+B(q,q) +˙ G(q) +CT(q)Λ =T(q,q),˙ (4.10) where M(q) is symmetric, positive definite and invertibly massive matrix, B(q,q) is the column vector˙ consisting of Coriolis forces, and G(q) is the column vector of gravitational forces acting on the system.
Non-Conservative Forces and the Power Dissipation Method
Impose no-slip conditions on the wheels via non-holonomic constraints and the Lagrange multipliers are given by equation 4.13. In summary, the dynamics of a wheeled, non-skid mobile robot at all its contact points can be described by equation 4.11. However, if the force at any wheel violates Equation 4.17, the PDP can be used to determine the exact slip condition,j, and the condition will evolve according to the pattern given by Equation 4.19.
Application to Axel
Remember that the Lagrangian function is given by L(q,q) =˙ K(q,q)˙ −V(q), where K(q,q) is the˙ kinetic energy and V(q) is the potential energy of the system. The velocity of the left wheel, v1, is calculated using the equations of rigid body kinematics:. For the holonomic constraint that constrains the tip of the control arm to lie above the ground plane. 4.27).
Modeling Extreme Terrain
The model switches between slip and no-slip states based on Coulomb's law of friction and the principle of power dissipation. For example, the headland in Victoria Crater, shown in Figure 1.4, can be divided into a relatively small number of non-mooring areas and areas that require mooring. By modeling the crater in this way, the general problem of motion planning on a complex 3-dimensional surface is transformed into a quasi-2-dimensional problem.
Obstacle Detection
Letαi,i= 1, .., N are the slope angles of the cable-demand and cable-free surfaces for a model consisting of N intersecting surfaces. Modifying Figure 4.4 so that the vertical axis of the truncated cones is at an angle to the vertical redefines compatibility for points on sloping terrain. Figures 4.5 and 4.6 show two different parts of the site, along with the output of the algorithm.
Algorithm
- Summary of the Problem
- Ascent/Descent Approach
- Ascent Path Planning
- Homotopies of Ascending Paths
- Constructing Anchor Reachable Sets
In practice, border edges form the boundaries of the regions where there is cable demand, or the edges of border obstacles. These covers represent the four unique homotopy classes of paths connecting the top-of-the-tether query region to the robot configuration. The sleeve of the red path is constructed by going up the tree from the end face to the start face (red marker).
Simple Example
A similar analysis can be used during descent planning to reduce the sleeve geometry to a subset of safe and controllable rover configurations. Then calculate the BTM of the tie-down demand planes and find the path inside the climb arm based on the optimization criterion, in this case the shortest path (Figure 4.15). Instead, the rover can find a safe path to the target and back to the anchor point by exploiting transient anchor points.
Shackleton Crater
Anchor available sets corresponding to smaller and larger values of the friction coefficient, µ, can be seen in figure 4.21. When the rover has less traction (smaller µ), the anchors that can be reached become narrower (Figure 4.21a), while more traction gives Axel access to a larger subset of the sleeve (Figure 4.21b). Yellow "holes" in the red area delineate the negative obstacles that are not part of the reachable anchors.
Tension on Steep Slopes
Understanding the tie-down forces while maneuvering on sloped terrain is therefore essential to ensure the safety of the rover. The figure also shows that the chain tension will increase dramatically at the ends of the swing arm angle, θc. The experiment suggests that Equation 5.1 can be used to predict cable tension during various maneuvers on steep slopes.
Climbing Precipices
The theoretical stress then quickly peaks as the wheels get closer to the corner of the ridge, reaching up to five times the weight of the rover (Figure 5.5a). Therefore, the end of the simulation represents the worst cable tension while climbing over a ridge. The potentially dangerous voltage spike can be mitigated by increasing the height of the anchor point.
Tension in the Field
The numbers in parentheses on the voltage graph correspond to the images in the time-lapse sequence. I agree with the conclusion of the analysis in section 5.2 that grooved wheels played a key role in the ascent of the chasm.
Axel 1 Shortcomings
If the cable is pulled out at low tension, slack cable will build up between Axel's body and the caster arm. Conversely, winding at a very high tension will cause the cable to cut through the lower layers of the wound cable, which can result in damage to the cable and make unwinding difficult. Additionally, there is no mechanism to align the chain as it rolls around the body.
Second-Generation Axel Prototype
The holes in the cams reduce wheel mass without compromising the overall strength of the wheel structure. The space within the volume of the wheel structure houses scientific instruments and sampling equipment. Additionally, double or triple redundant communications lines embedded in the cable ensure that the rover's operators will never lose contact with Axel 2.
The DuAxel Architecture
In this mode, the central module and one Axel form the mothership and physical anchor, enabling the second Axel to jump into low ground (Figure 6.4). In the third and final configuration, the central module is deposited by the Axels and serves as a fixed mothership while the Axel 2 robots explore the surrounding area. By coordinating tasks between the two rovers, DuAxel can quickly and efficiently study a specific location of interest, making full use of the central module's sophisticated scientific instruments.
Extreme Terrain Field Tests
- Vulcan Quarry
- Arizona Desert
At the bottom of the traverse, a power supply connected to the chain was used to remotely charge Axel from the top of the drop. DuAxel followed the rocky terrain on the side (Figure 6.7) to climb to the top of the hill and support Axel as he charged down the steep slope (Figure 6.8). Additional ankle scans were performed at each of the two sites used during the second day of testing.
Future Work
The approach used in all previous experiments was open loop, that is, the tether was unrolled at a constant rate while Axel maneuvered the terrain. With these simplifications, the tight tether takes the form of the shortest homotopic path, which is simply the shortest path in the given sleeve of the tether. Autonomous planning in the field will likely need to incorporate a combination of sensors to detect nuances in the terrain and adjust to the tether configuration.
