Yoshiaki Kuwata - Academia.edu (original) (raw)

Papers by Yoshiaki Kuwata

The Proceedings of the Transportation and Logistics Conference

This paper considers automatic emergency recovery control system of an accident aircraft． This sy... more This paper considers automatic emergency recovery control system of an accident aircraft． This syslem is divided into three par1s ， 1) Neural Network identifies deterioration of control lnputs and bias forces due to accident or failure ， 2) the distribution matrix which distributes required control forces to control inp 山 s are adjusted by using the identified control effectiveness ， and 3)the numerical optimizer modifies the reference inputs of flight controller ． Numerical Examp 【 e demonstrates the control process of an aircraft with unexpected actuator failure and indicates the importance of artificial control inputs for Iearning the Neural Network ．

... 1.1.1 Model Predictive Control/Receding Horizon Control Trajectory planning for autonomous ve... more ... 1.1.1 Model Predictive Control/Receding Horizon Control Trajectory planning for autonomous vehicles has been studied in many fields including ... the system to predict its future behavior and generate control inputs that satisfy the ... so the algorithm using MPC is explicitly adaptive. ...

This set of notes builds on the previous two chapters and explores the use of online optimization... more This set of notes builds on the previous two chapters and explores the use of online optimization as a tool for control of nonlinear control. We begin with a high-level discussion of optimization-based control, refining some of the concepts initially introduced in Chapter 1. We then describe the technique of receding horizon control (RHC), including a proof of stability for a particular form of receding horizon control that makes use of a control Lyapunov function as a terminal cost. We conclude the chapter with a detailed design example, in which we can explore some of the computational tradeoffs in optimization-based control. Prerequisites. Readers should be familiar with the concepts of trajectory generation and optimal control as described in Chapters 1 and ??. For the proof of stability for the receding horizon controller that we present, familiarity with Lyapunov stability analysis at the level given inÅM08, Chapter 4 (Dynamic Behavior) is assumed. The material in this chapter is based on part on joint work with John Hauser and Ali Jadbabaie [MHJ + 03]. 3.1 Optimization-Based Control Optimization-based control refers to the use of online, optimal trajectory generation as a part of the feedback stabilization of a (typically nonlinear) system. The basic idea is to use a receding horizon control technique: a (optimal) feasible trajectory is computed from the current position to the desired position over a finite time T horizon, used for a short period of time δ < T , and then recomputed based on the new system state starting at time t + δ until time t + T + δ. Development and application of receding horizon control (also called model predictive control, or MPC) originated in process control industries where the processes being controlled are often sufficiently slow to permit its implementation. An overview of the evolution of commercially available MPC technology is given in [QB97] and a survey of the state of stability theory of MPC is given in [MRRS00]. Design approach The basic philosophy that we propose is illustrated in Figure 3.1. We begin with a nonlinear system, including a description of the constraint set. We linearize this system about a representative equilibrium point and perform a linear control design using standard control design tools. Such a design can provide provably robust performance around the equilibrium point and, more importantly, allows the designer 3.1. OPTIMIZATION-BASED CONTROL

Advances in the Astronautical Sciences, 2004

This paper introduces two unique testbeds that have recently been developed at MIT to demonstrate... more This paper introduces two unique testbeds that have recently been developed at MIT to demonstrate the coordination and control of teams of multiple autonomous vehicles. The first testbed uses eight rovers and four blimps operated indoors to emulate a heterogeneous fleet of vehicles that could be used to perform a search and rescue mission. The second testbed uses eight 60-scale model aircraft that are flown autonomously using a commercially available autopilot. This combination of testbeds provides platforms for both advanced ...

We discuss the development, integration, simulation, and flight test of a manned vehicle-UAV miss... more We discuss the development, integration, simulation, and flight test of a manned vehicle-UAV mission system in a partially-known environment. The full control system allows a manned aircraft to issue mission level commands to an autonomous aircraft in realtime. This system includes a Natural Language (NL) Interface to allow the manned and unmanned vehicle to communicate in languages understood by both agents. The unmanned vehicle implements a dynamic mission plan determined by a weapons systems officer (WSO) on ...

Proceedings of Spie the International Society For Optical Engineering, Apr 13, 2009

This paper considers a landing problem for an MAV that uses only a monocular camera for guidance.... more This paper considers a landing problem for an MAV that uses only a monocular camera for guidance. Although this sensor cannot measure the absolute distance to the target, by using optical flow algorithms, time-to-collision to the target is obtained. Existing work has applied a simple proportional feedback control to simple dynamics and demonstrated its potential. However, due to the singularity in the time-to-collision measurement around the target, this feedback could require an infinite control action. This paper extends the approach into nonlinear dynamics. In particular, we explicitly consider the saturation of the actuator and include the effect of the aerial drag. It is shown that the convergence to the target is guaranteed from a set of initial conditions, and the boundaries of such initial conditions in the state space are numerically obtained. The paper then introduces parametric uncertainties in the vehicle model and in the time-to-collision measurements. Using an argument similar to the nominal case, the robust convergence to the target is proven, but the region of attraction is shown to shrink due to the existence of uncertainties. The numerical simulation validates these theoretical results.

AIAA Guidance, Navigation, and Control Conference and Exhibit, 2004

This paper presents a receding horizon controller (RHC) that can be used to design trajectories f... more This paper presents a receding horizon controller (RHC) that can be used to design trajectories for an aerial vehicle flying through a three dimensional terrain with obstacles and no-fly zones. To avoid exposure to threats, the paths are chosen to stay as close to the terrain as possible, but the vehicle can choose to pop-up over the obstacles if necessary. The approach is similar to our previous two-dimensional algorithms that construct a coarse cost map to provide approximate paths from a sparse set of nodes to the goal and then use Mixed-integer Linear Programming (MILP) optimization to design a detailed trajectory. The main contribution of this paper is to extend this approach to 3D, in particular providing a new algorithm for connecting the cost map and the detailed path in the MILP. This connection is done by introducing a new cost-to-go function that includes an altitude penalty and accounts for the vehicle dynamics. Initial guess for MILP RHC is constructed from the previous solution and is shown to reduce the solution time. Several simulation results are presented to show that the path planning algorithm yields good overall performance and is computationally tractable in a complex environment.

Proceedings of the 2003 American Control Conference, 2003., 2003

This paper describes methods for optimizing the task allocation problem for a fleet of Unmanned A... more This paper describes methods for optimizing the task allocation problem for a fleet of Unmanned Aerial Vehicles (UAVs) with tightly coupled tasks and rigid relative timing constraints. The overall objective is to minimize the mission completion time for the fleet, and the task assignment must account for differing UAV capabilities and no-fly zones. Loitering times are included as extra degrees of freedom in the problem to help meet the timing constraints. The overall problem is formulated using Mixed-integer Linear Programming (MILP), which gives the globally optimal solution. An approximate decomposition solution method is also used to overcome the computational issues that arise when using MILP for larger problems. The problem is also posed in a way that can be solved using Tabu search. This approach is demonstrated to provide good solutions in reasonable computation times for large problems that are very difficult to solve using the exact or approximate decomposition methods.

Advances in Computer Vision and Pattern Recognition, 2014

Unmanned Systems Technology XVI, 2014

AIAA SPACE 2010 Conference & Exposition, 2010

The Jet Propulsion Laboratory is developing the Lunar Surface Operations Simulator software packa... more The Jet Propulsion Laboratory is developing the Lunar Surface Operations Simulator software package to support analyses for future NASA lunar missions. The package is built on and extended from previous simulation packages developed at JPL. It simulates mechanical motion, soil interaction, environmental, and physical processes. Physical process dynamics include environmental control and life support, thermal, radiation and power transients. An integrated architecture allows use of common models and enables interactions between components operating in different domains to be easily modeled. We describe recent developments and analyses performed to support lunar surface missions and analog field trials.

This paper discusses several path planning challenges for planetary robots and the ongoing effort... more This paper discusses several path planning challenges for planetary robots and the ongoing efforts to address them. The considered challenges include limited onboard computing resources, degraded/failed mechanical parts, and specific operating conditions such as the influence of atmospheric currents on aerobots with significant dynamics. These are examples of real-world challenges identified in the current and potential future missions. For each of the challenges, the paper also illustrates potential future solutions from the current research efforts.

American Control Conference, 2004. …, Jun 30, 2004

This work presents a formulation of a stable receding horizon controller (RHC) for the minimum ti... more This work presents a formulation of a stable receding horizon controller (RHC) for the minimum time trajectory optimization problem with a vehicle flying in a complex environment with obstacles and no-fly zones. The overall problem is formulated using mixed-integer linear programming (MILP). The RHC uses a simple vehicle dynamics model in the near term and an approximate path model in the long term. This combination gives a good estimate of the cost-to-go and greatly reduces the computational effort required to design ...

AIAA guidance, navigation, and control conference, Monterey, CA, Aug 5, 2002

This paper addresses the problems of autonomous task allocation and trajectory planning for a fle... more This paper addresses the problems of autonomous task allocation and trajectory planning for a fleet of UAVs. Two methods are compared for solving the optimization that combines task assignment, subjected to UAV capability constraints, and path planning, subjected to dynamics, avoidance and timing constraints. Both sub-problems are non-convex and the two are strongly-coupled. The first method expresses the entire problem as a single mixed-integer linear program (MILP) that can be solved using ...

… . Proceedings of the …, Jun 30, 2004

This paper introduces two unique testbeds that have recently been developed to demonstrate the co... more This paper introduces two unique testbeds that have recently been developed to demonstrate the cooperative control of teams of UAVs. The first testbed uses eight rovers and four blimps operated indoors to emulate a team of heterogeneous vehicles performing a combined reconnaissance and strike mission. The second testbed uses eight small aircraft that are flown autonomously using a commercially available autopilot. This combination of testbeds provides platforms for both advanced research and realistic demonstrations. ...

2010 IEEE Aerospace Conference, 2010

The recent Titan Saturn System Mission (TSSM) proposal incorporates a montgolfière (hot-air ballo... more The recent Titan Saturn System Mission (TSSM) proposal incorporates a montgolfière (hot-air balloon) as part of its architecture. The authors have conducted a study to determine the impact of using the Titan wind field to extend the scientific reach of the balloon. The results show that a windassisted unpropelled montgolfière will be able to reach a broad set of science targets, while a windassisted propelled montgolfière could reach any area of interest on Titan, and do so in a fraction of the time needed by the unpropelled balloon.

This thesis investigates the coordination and control of fleets of unmanned aerial vehicles (UAVs... more This thesis investigates the coordination and control of fleets of unmanned aerial vehicles (UAVs). Future UAVs will operate autonomously, and their control systems must compensate for significant dynamic uncertainty. A hierarchical approach has been proposed to account for various types of uncertainty at different levels of the control system. The resulting controller includes task assignment, graph-based coarse path planning, detailed trajectory optimization using receding horizon control (RHC), and a low-level waypoint follower. Mixed-integer linear programming (MILP) is applied to both the task allocation and trajectory design problems to encode logical constraints and discrete decisions together with the continuous vehicle dynamics.