【论文精读】- 《Translational and Scaling Formation Maneuver Control via a Bearing-Based Approach》

Posted by Orchid on March 11, 2025
  • 方法:bearing-based approach
    • 基于方位的编队控制,利用智能体之间相对方位信息来实现目标编队形状的维持和控制(方位即方向)
  • 编队形状定义:inter-neighbor bearings
  • 实现效果:Translational and Scaling

过去的一些方法

  1. The problem of formation scale control has been studied by the relative-position and distance-based approaches

    • 缺陷:
    • 当编队伸缩时,relative-position and distance 是变化的,每个 follower 需要估计由 leaders 确定的期望 scaling
    • 这两种方法在以往研究,the desired formation scale is constant
  2. complex Laplacian matrix

    相对完善

    the target formation is defined by complex linear constraints that are invariant to the translation, rotation, and scale of the formation

    • 缺陷:停留在二维,想要扩展比较难

贡献

  1. 研究的前提条件:当目标队形能由 inter-neighbor bearings and leader agents 唯一确定

    image-20250310213618526

  2. 使用了一种特殊的矩阵 bearing Laplacian,来表征 the interconnection topology and the inter-neighbor bearings of the formation

  3. 提出了相应的两条线性的控制律(需要不同的信息输入),针对 double-integrator dynamics(输入加速度,输出位置)。注意,只有 leaders 知道 the desired translational and scaling maneuver,followers 只知道相邻 agents 的信息(相对位置、相对速度)

  4. 上述控制律能够应对恒定输入干扰和加速度饱和,并分析了其稳定性

Problem Formulation

\[\begin{align} g_{ij} &\triangleq \frac{p_j - p_i}{\|p_j - p_i\|}, \\ &= \begin{bmatrix} \Delta X \\ \Delta Y \end{bmatrix} \\ P_{g_{ij}} &\triangleq I_d - g_{ij} g_{ij}^T. \\ &=\begin{bmatrix} 1-\Delta X^2 & \Delta X\Delta Y \\ \Delta X\Delta Y & 1-\Delta Y^2 \end{bmatrix} \end{align}\]

Bearing-Based Formation Maneuver Control

The target formation denoted by $ G(p^*(t)) $ is a formation that satisfies the following constraints for all $ t \geq 0 $:

\[\begin{align} &(a) Bearing:\ \frac{(p_j^*(t) - p_i^*(t))}{\|p_j^*(t) - p_i^*(t)\|} = g_{ij}^*, \forall (i, j) \in \mathcal{E} \\ &(b) Leader:\ p_i^*(t) = p_i(t), \forall i \in \mathcal{V}_\ell \end{align}\]

define the position and velocity errors for the followers as:

\[\delta_p(t) = p_f(t) - p_f^*(t), \quad \delta_v(t) = v_f(t) - v_f^*(t)\]

Properties of the Target Formation

  • Bearing Laplacian Matrix: Define a matrix $ \mathcal{B}(G(p^*)) \in \mathbb{R}^{dn \times dn} $ with the $ ij $-th block of submatrix as

    \[[\mathcal{B}(G(p^*))]_{ij} = \begin{cases} \mathbf{0}_{d \times d}, & i \neq j, (i, j) \notin \mathcal{E}, \\ -P_{g_{ij}^*}, & i \neq j, (i, j) \in \mathcal{E}, \\ \sum_{k \in \mathcal{N}_i} P_{g_{ik}^*}, & i = j, i \in \mathcal{V}. \end{cases}\]

    该矩阵同时表征了 the interconnection topology and the bearings of the formation.

    • 性质一:

      \[\text{Null}(\mathcal{B}) \supseteq \text{span}\{ \mathbf{1}_n \otimes I_d, p^* \} \\\]

      证明很简单:

      \[\mathcal{B}x = \left[ \begin{array}{c} \vdots \\ \sum_{j \in \mathcal{N}_i} P_{g_{ij}^*} (x_i - x_j) \\ \vdots \end{array} \right].\]
    • 性质二:

      \[\mathcal{B} = \left[ \begin{array}{cc} \mathcal{B}_{\ell\ell} & \mathcal{B}_{\ell f} \\ \mathcal{B}_{f\ell} & \mathcal{B}_{ff} \end{array} \right]\]

      $\mathcal{B}_{ff} \in \mathbb{R}^{dn_f \times dn_f}$ 是对称且半对称的

  • Uniqueness of the Target Formation

    • 定理一: 当给定可行的 bearing constraints and leader positions,目标编队可以唯一确定,当且仅当 $\mathcal{B}_{ff}$ 非奇异。并且 followers 的目标位置和速度可以确定:

      \[\begin{align*} p^*_f(t) &= -\mathcal{B}_{ff}^{-1} \mathcal{B}_{f\ell} p_{\ell}(t), \\ v^*_f(t) &= -\mathcal{B}_{ff}^{-1} \mathcal{B}_{f\ell} v_{\ell}(t). \end{align*}\]

      证明:

      由性质一:

      \[\mathcal{B} p^* = 0\]

      结合性质二:

      \[\mathcal{B}_{ff} p_f^* + \mathcal{B}_{f\ell} p_{\ell} = 0\]

      当 $\mathcal{B}_{ff}$ 非奇异:

      \[p_f^* = -\mathcal{B}_{ff}^{-1} \mathcal{B}_{f\ell} p_{\ell}\\ v_f^* = \dot{p}_f^* = -\mathcal{B}_{ff}^{-1} \mathcal{B}_{f\ell} v_{\ell}\]

      也就是说,目标编队的位置和速度可以由 leaders 唯一确定。

    • 其他论文给出的结论:A useful sufficient condition is that the target formation is unique if it is infinitesimally bearing rigid and has at least two leaders

  • Target Formation Maneuvering

    这一部分定义了目标编队的运动,包括其平移和缩放运动

    首先,定义了目标编队的中心以及规模(centroid and scale):

    \[\begin{align} c(p^*(t)) &\triangleq \frac{1}{n} \sum_{i \in \mathcal{V}} p_i^*(t) = \frac{1}{n} (\mathbf{1}_n \otimes I_d)^T p^*(t)\\ s(p^*(t)) &\triangleq \sqrt{\frac{1}{n} \sum_{i \in \mathcal{V}} \|p_i^*(t) - c(p^*(t))\|^2}\\ &= \frac{1}{\sqrt{n}} \|p^*(t) - \mathbf{1}_n \otimes c(p^*(t))\| \end{align}\]

    接着,定义了 the desired maneuvering dynamics of the centroid and scale of the target formation:

    \[\dot{c}(p^*(t)) = v_c(t), \quad \dot{s}(p^*(t)) = \alpha(t) s(p^*(t))\]
    • 定理二:(比较显然)当实现目标编队的运动时,leaders 的速度应是以下形式:

      \[v_i(t) = v_c(t) + \alpha(t) [p_i(t) - c(p^*(t))], \quad i \in \mathcal{V}_\ell\]

Bearing-based formation control laws

这一小节提出了两种控制律(针对 followers)。第一种控制律,是 based on Constant Leader Velocity,需要相对位置和相对速度信息;第二种控制律,是 based on Time-Varying Leader Velocity,需要相对位置、相对速度以及加速度信息。

  • 控制律 1:Constant Leader Velocity

    \[u_i = -\sum_{j \in \mathcal{N}_i} P_{g_{ij}^*} \left[ k_p (p_i - p_j) + k_v (v_i - v_j) \right]\]
  • 控制律 2:Time-Varying Leader Velocity

    \[u_i = -K_i^{-1} \sum_{j \in \mathcal{N}_i} P_{g_{ij}^*} \left[ k_p (p_i - p_j) + k_v (v_i - v_j) - \dot{v}_j \right]\]

文中给出了两条控制律收敛性的证明。

Bearing-based formation control with practical issues

  • Constant Input Disturbance 输入噪声干扰

    假设 followers 的控制输入存在恒定的噪声,即:

    \(\dot{p}_i = v_i, \quad \dot{v}_i = u_i + \mathbf{w}_i\) 针对上述的两条控制律,引入积分项,分别给出了修正:

    • 控制律 1:Constant Leader Velocity(修正)

      \[u_i = -\sum_{j \in \mathcal{N}_i} P_{g_{ij}^*} \left[ k_p (p_i - p_j) + k_v (v_i - v_j) + k_I \int_0^t (p_i - p_j) \mathrm{d}\tau \right]\]
    • 控制律 2:Time-Varying Leader Velocity(修正)

      \[u_i = -K_i^{-1} \sum_{j \in \mathcal{N}_i} P_{g_{ij}^*} \left[ k_p (p_i - p_j) + k_v (v_i - v_j) - \dot{v}_j + k_I \int_0^t (p_i - p_j) \mathrm{d}\tau \right]\]

    给出了条件限制:

    \[0 < k_I < k_p k_v \lambda_{\min}(\mathcal{B}_{ff})\]

    并给出了收敛性证明