Publications Search Results

Now showing 1 - 8 of 8
  • Publication
    Robot spray painting trajectory optimization
    ( 2020)
    Gleeson, D.
    ;
    Jakobsson, S.
    ;
    Salman, R.
    ;
    Sandgren, N.
    ;
    Edelvik, F.
    ;
    Carlson, J.S.
    ;
    Lennartson, B.
    In the manufacturing industry, spray painting is often an important part of the process. Especially in the automotive industry, the perceived quality of the final product is closely linked to the exactness and smoothness of the painting process. For complex products or low batch size production, manual spray painting is often used. But in large scale production with a high degree of automation, the painting is normally performed by industrial robots. There is a need to improve and simplify the generation of robot trajectories used in industrial paint booths. A method for spray paint optimization is presented, which can be used to smooth out an initial trajectory and minimize paint thickness deviations from a target thickness. By fitting a spline function to experimental data, an applicator footprint profile is determined, which is a two-dimensional reference function of the applied paint thickness. This footprint profile is then projected to the geometry and used as a d eposition model at each point along the trajectory. The positions and durations of all trajectory segments are used as optimization variables. They are modified with the primary goal to obtain a paint applicator trajectory, which will closely match a target paint thickness when executed. The algorithm is shown to produce satisfactory results on both a simple 2-dimensional test example, and a nontrivial industrial case of painting a tractor render. The final trajectory shows an overall thickness close to the target thickness, and the resulting trajectory is feasible to execute directly on an industrial robot.
  • Publication
    Inverse dynamics for discrete geometric mechanics of multibody systems with application to direct optimal control
    ( 2018)
    Björkenstam, S.
    ;
    Leyendecker, S.
    ;
    Linn, J.
    ;
    Carlson, J.
    ;
    Lennartson, B.
    In this paper, we present efficient algorithms for computation of the residual of the constrained discrete Euler-Lagrange (DEL) equations of motion for tree structured, rigid multibody systems. In particular, we present new recursive formulas for computing partial derivatives of the kinetic energy. This enables us to solve the inverse dynamics problem of the discrete system with linear computational complexity. The resulting algorithms are easy to implement and can naturally be applied to a very broad class of multibody systems by imposing constraints on the coordinates by means of Lagrange multipliers. A comparison is made with an existing software package, which shows a drastic improvement in computational efficiency. Our interest in inverse dynamics is primarily to apply direct transcription optimal control methods to multibody systems. As an example application, we present a digital human motion planning problem, which we solve using the proposed method. Furthermore, we present detailed descriptions of several common joints, in particular singularity-free models of the spherical joint and the rigid body joint, using the Lie groups of unit quaternions and unit dual quaternions, respectively.
  • Publication
    Reinforcement learning in real-time geometry assurance
    ( 2018)
    Jorge, E.
    ;
    Brynte, L.
    ;
    Cronrath, C.
    ;
    Wigström, O.
    ;
    Bengtsson, K.
    ;
    Gustavsson, E.
    ;
    Lennartson, B.
    ;
    Jirstrand, M.
    To improve the assembly quality during production, expert systems are often used. These experts typically use a system model as a basis for identifying improvements. However, since a model uses approximate dynamics or imperfect parameters, the expert advice is bound to be biased. This paper presents a reinforcement learning agent that can identify and limit systematic errors of an expert systems used for geometry assurance. By observing the resulting assembly quality over time, and understanding how different decisions affect the quality, the agent learns when and how to override the biased advice from the expert software.
  • Publication
    Towards energy optimization using trajectory smoothing and automatic code generation for robotic assembly
    ( 2016)
    Gleeson, D.
    ;
    Björkenstam, S.
    ;
    Bohlin, R.
    ;
    Carlson, J.S.
    ;
    Lennartson, B.
    In automated industrial production, the efficiency of robotic motions directly affects both the final throughput and the energy consumption. By simulating and optimizing robot trajectories, cycle times and energy consumption can be lowered, or redundant robots can be detected. Here a polynomial basis function trajectory parametrization is presented, which enables direct export to executable robot code, and reduces the number of variables in the optimization problem. The algorithm finds time-optimal trajectories, while including collision avoidance and fulfilling joint, velocity and acceleration limitations. Applied torques are used as an approximation of the energy consumption to analyse the smooth trajectories, and successful tests show potential reductions of 10% for a standard industrial robot stud welding station.
  • Publication
    Enhancing digital human motion planning of assembly tasks through dynamics and optimal control
    ( 2016)
    Björkenstam, S.
    ;
    Delfs, N.
    ;
    Carlson, J.S.
    ;
    Bohlin, R.
    ;
    Lennartson, B.
    Better operator ergonomics in assembly plants reduce work related injuries, improve quality, productivity and reduce cost. In this paper we investigate the importance of modeling dynamics when planning for manual assembly operations. We propose modeling the dynamical human motion planning problem using the Discrete Mechanics and Optimal Control (DMOC) method, which makes it possible to optimize with respect to very general objectives. First, two industrial cases are simulated using a quasi-static inverse kinematics solver, demonstrating problems where this approach is sufficient. Then, the DMOC-method is used to solve for optimal trajectories of a lifting operation with dynamics. The resulting trajectories are compared to a steady state solution along the same path, indicating the importance of using dynamics.
  • Publication
    Optimizing robot trajectories for automatic robot code generation
    ( 2015)
    Gleeson, D.
    ;
    Björkenstam, S.
    ;
    Bohlin, R.
    ;
    Carlson, J.S.
    ;
    Lennartson, B.
    In manufacturing industry, the automation of robot trajectory planning and robot code generation lowers the development and assessment times of new products and models, increasing the flexibility of a robotic production line. This is of increasing importance in a number of industries where new products and new models are constantly being developed. This paper presents a method that automatically generates robot code for collision free trajectories. Starting with a collision free piecewise linear joint space trajectory, a direct transcription optimal control method is used, iteratively, to improve the trajectory while maintaining a desired clearance. The final solution is a collision free robot trajectory defined by a set of parameters, e.g. joint space via points and TCP-zone radii, that can be used to automatically generate functional RAPID code for an industrial robot. The algorithm is designed to be easily extendable. For example, by changing the constraints on the trajectory or by including energy consumption as an objective. A successful test of the algorithm has been carried out on an industrial stud welding station using the ABB virtual controller, indicating a 10% improvement in motion time.
  • Publication
    Exploiting sparsity in the discrete mechanics and optimal control method with application to human motion planning
    ( 2015)
    Björkenstam, S.
    ;
    Carlson, J.S.
    ;
    Lennartson, B.
    The discrete equations of motion derived using a variational principle are particularly attractive to be used in numerical optimal control methods. This is mainly because: i) they exhibit excellent energy behavior, ii) they extend gracefully to systems with holonomic constraints and iii) they admit compact representation of the discrete state space. In this paper we propose the use of sparse finite differencing techniques for the Discrete Mechanics and Optimal Control method. In particular we show how to efficiently construct estimates of the Jacobian and Hessian matrices when the dynamics of the optimal control problem is discretized using a variational integrator. To demonstrate the effectiveness of this scheme we solve a human motion planning problem of an industrial assembly task, modeled as a multibody system consisting of more than one hundred degrees of freedom.
  • Publication
    Energy efficient and collision free motion of industrial robots using optimal control
    ( 2013)
    Björkenstam, S.
    ;
    Gleeson, D.
    ;
    Bohlin, R.
    ;
    Carlson, J.S.
    ;
    Lennartson, B.
    In a production plant for complex assembled products there could be up to several hundred of robots used for handling and joining operations. Thus, improvement in robot motions can have a huge impact on equipment utilization and energy consumption. These are two of the most important aspects of sustainability in a production system. Therefore, this paper presents an algorithm for generating efficient and collision free motion of industrial robots using path planning and direct transcription methods for numerical optimal control. As a measure of efficiency for moving between configurations we use a combination of the energy norm of the applied actuator torques and the cycle time. Velocity and torque limits are handled and modeled as hard constraints. However, more general problems can be solved by the same approach. Our novel algorithm solves the problem in three steps; (i) first a path planning algorithm calculates an initial collision free path, (ii) a convex optimal c ontrol problem is then formulated to follow this path, and finally (iii) a nonlinear optimal control problem is solved to iteratively improve the trajectory. The resulting trajectory is guaranteed to be collision free by restrictions in the configuration space based on a local sensitivity analysis. The algorithm has been successfully applied to several industrial cases demonstrating that the proposed method can be used effectively in practical applications.