toplogo
Sign In

Optimal Trajectory Generation for Autonomous On-Orbit Assembly with Non-Conservative Collision Avoidance


Core Concepts
This paper presents a novel non-conservative collision avoidance technique using convex optimization to establish the distance between robotic spacecraft and space structures for autonomous on-orbit assembly operations.
Abstract
The paper proposes a novel approach to reformulate the collision avoidance problem by precisely modeling complex-shaped, full-dimensional controlled vehicles and obstacles using a series of differentiable convex functions. These functions are used as constraints in a convex optimization problem to establish the minimum distance between any two convex sets. The optimality condition of this optimization problem forms a new set of differentiable constraints that can be used in an optimal control problem to generate collision-free trajectories. The effectiveness of the proposed method is demonstrated in two autonomous on-orbit assembly scenarios in tight environments, where the robotic spacecraft performs the assembly procedure. A pseudo-spectral optimal control method is utilized to show that the proposed technique can generate optimal trajectories in tight environments with multiple active components present. The key highlights and insights are: The controlled vehicle and obstacles are modeled using real-valued, differentiable, convex functions, which are then used in a convex optimization problem to establish the minimum distance between them. The optimality conditions of this convex optimization problem are used to formulate a new set of differentiable constraints that can be incorporated into an optimal control problem to enforce collision avoidance. The proposed method is non-conservative and can handle full-dimensional controlled vehicles and obstacles, making it suitable for autonomous on-orbit assembly operations in tight environments. Numerical simulations of two assembly scenarios demonstrate the effectiveness of the proposed technique in generating optimal, collision-free trajectories for the robotic spacecraft.
Stats
The maximum available thrust and torque were selected as 0.02 N and 0.01 N.m, respectively. The robotic spacecraft unit's mass and inertia were considered 3 kg and 5 kg/m^2, respectively. 20 nodes were used for discretization in Scenario A.
Quotes
"The collision avoidance constraints are prominent as non-convex, non-differentiable, and challenging when defined in optimization-based motion planning problems." "The necessity of proposing a collision avoidance technique that is not conservative in any manner, is designed for full-dimensional spacecraft and obstacles, and is comprehensive to incorporate the space robotics applications, as well as other applications, is evident."

Deeper Inquiries

How can the proposed technique be extended to handle dynamic obstacles or moving target structures during the autonomous on-orbit assembly process

The proposed technique can be extended to handle dynamic obstacles or moving target structures during the autonomous on-orbit assembly process by incorporating predictive modeling and real-time updates. Predictive Modeling: By utilizing predictive algorithms, the system can anticipate the movement of dynamic obstacles or target structures based on historical data or real-time sensor inputs. This information can be used to adjust the trajectory of the robotic spacecraft to avoid collisions proactively. Real-Time Updates: Implementing a feedback loop that continuously updates the collision avoidance constraints based on the changing positions and velocities of the dynamic obstacles or target structures. This real-time information can be integrated into the optimization problem to generate collision-free trajectories. Dynamic Constraints: Introducing dynamic constraints that account for the motion of the obstacles or target structures in the optimization problem. These constraints can be formulated to ensure a safe distance is maintained between the robotic spacecraft and the moving objects at all times. Collision Prediction: Utilizing collision prediction algorithms to forecast potential collisions between the robotic spacecraft and dynamic obstacles. By predicting these collisions in advance, the system can take preemptive actions to avoid them.

What are the potential limitations or drawbacks of the convex modeling approach used in this paper, and how can they be addressed

The convex modeling approach used in the paper has several potential limitations and drawbacks that need to be addressed: Conservativeness: One limitation of convex modeling is its inherent conservatism, which may result in overly cautious trajectories that sacrifice efficiency. This can be addressed by refining the modeling techniques to better capture the actual shapes and dynamics of the objects involved. Complexity: Convex modeling may oversimplify the shapes of the objects, leading to inaccuracies in the collision avoidance constraints. To address this, more sophisticated modeling techniques, such as using higher-order polynomials or splines, can be employed to better represent the objects' geometry. Computational Efficiency: Convex optimization problems can become computationally intensive, especially as the number of constraints and variables increases. To mitigate this, optimization algorithms can be optimized, and parallel computing techniques can be utilized to improve efficiency. Real-Time Adaptability: Convex modeling may struggle to adapt to real-time changes in the environment or object dynamics. Implementing adaptive algorithms that can update the constraints on the fly based on new information can help overcome this limitation.

How can the proposed collision avoidance framework be integrated with other components of the autonomous on-orbit assembly system, such as task allocation and robotic manipulation, to enable a complete autonomous solution

To integrate the proposed collision avoidance framework with other components of the autonomous on-orbit assembly system, such as task allocation and robotic manipulation, a holistic approach is required: Task Allocation: The collision avoidance constraints should be incorporated into the task allocation system to ensure that the planned assembly tasks consider the constraints from the collision avoidance framework. This integration will enable the system to generate feasible and collision-free task schedules. Robotic Manipulation: The collision avoidance framework should be integrated with the robotic manipulation algorithms to ensure that the robotic spacecraft's movements are coordinated with the collision avoidance constraints. This integration will enable the robotic manipulators to adjust their actions in real-time to avoid collisions. Sensor Fusion: Integrating sensor data from the robotic spacecraft with the collision avoidance framework can enhance the system's awareness of its surroundings. By fusing sensor data with the collision avoidance constraints, the system can make more informed decisions and adapt to dynamic environments. Feedback Loop: Establishing a feedback loop between the collision avoidance framework and the overall autonomous on-orbit assembly system will enable continuous improvement and adaptation. The system can learn from past experiences and optimize its performance over time. By integrating the collision avoidance framework with other components of the autonomous on-orbit assembly system, a comprehensive and efficient autonomous solution can be achieved.
0