Geometric Algebra for Optimal Control with Applications in Manipulation Tasks

Tobias Löw,Sylvain Calinon

DOI: https://doi.org/10.1109/TRO.2023.3277282

2023-06-05

Abstract:Many problems in robotics are fundamentally problems of geometry, which lead to an increased research effort in geometric methods for robotics in recent years. The results were algorithms using the various frameworks of screw theory, Lie algebra and dual quaternions. A unification and generalization of these popular formalisms can be found in geometric algebra. The aim of this paper is to showcase the capabilities of geometric algebra when applied to robot manipulation tasks. In particular the modelling of cost functions for optimal control can be done uniformly across different geometric primitives leading to a low symbolic complexity of the resulting expressions and a geometric intuitiveness. We demonstrate the usefulness, simplicity and computational efficiency of geometric algebra in several experiments using a Franka Emika robot. The presented algorithms were implemented in c++20 and resulted in the publicly available library \textit{gafro}. The benchmark shows faster computation of the kinematics than state-of-the-art robotics libraries.

Robotics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to use Geometric Algebra (GA) to optimize control problems in robotic manipulation tasks. Specifically, the paper shows how to use geometric algebra to unify and simplify the cost function modeling of different geometric primitives, thereby reducing symbolic complexity and enhancing geometric intuitiveness. Through several experiments, the paper demonstrates the effectiveness and computational efficiency of geometric algebra in handling robotic manipulation tasks. In particular, compared with existing advanced robotic libraries, geometric algebra can calculate kinematics more quickly.

### Core problems of the paper:
1. **Unifying geometric methods**: The paper proposes using geometric algebra to unify and generalize existing geometric methods (such as screw theory, Lie algebra, and dual quaternions) to solve geometric problems in robotic manipulation tasks.
2. **Cost function modeling**: The paper explores how to use geometric algebra to construct cost functions that can be consistently applied to different geometric primitives, thereby simplifying the modeling of optimal control problems.
3. **Computational efficiency**: The paper experimentally demonstrates the high efficiency of using geometric algebra in calculating kinematics and provides an open - source library `gafro` that implements all the algorithms in the paper.

### Specific contributions:
1. **Extended Lagrangian dynamics**: Extended the Lagrangian dynamics of serial manipulators in conformal geometric algebra, including non - trivial inertial tensors, and used it in an inverse - dynamics control scheme.
2. **Optimized objective function**: Proposed using geometric algebra to define the objective function in optimization problems, especially in inverse kinematics and optimal control problems, and showed how to model different geometric relationships while maintaining the consistency of the cost function structure.
3. **Integration of existing frameworks**: Showed how to seamlessly integrate geometric algebra formulas into existing frameworks based on linear matrix algebra, taking advantage of the sparsity of geometric algebra to promote its application.
4. **Open - source library**: Provided an open - source library named `gafro` that implements all the proposed formulas and algorithms. This library is based on a custom - implemented conformal geometric algebra and uses expression template technology. Benchmark tests show that it calculates kinematics faster than existing advanced robotic libraries.

### Conclusion:
By demonstrating the application of geometric algebra in robotic manipulation tasks, the paper not only solves the problem of how to unify and simplify geometric methods but also proves its high efficiency and effectiveness in practical applications. Through the provided open - source library `gafro`, researchers and engineers can more easily apply geometric algebra to robotic control systems.