193号资源-源程序：SCI文档+程序动态四足机器人的自由模型预测控制--------已提供下载资源

👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆下载资源链接👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆

《《《《《《《《更多资源还请持续关注本专栏》》》》》》》

论文与完整源程序_电网论文源程序的博客-CSDN博客https://blog.csdn.net/liang674027206/category_12531414.html

电网论文源程序-CSDN博客电网论文源程序擅长文章解读,论文与完整源程序,等方面的知识,电网论文源程序关注python,机器学习,计算机视觉,深度学习,神经网络,数据挖掘领域.https://blog.csdn.net/LIANG674027206?type=download

SCI《 Real-time Model Predictive Control for Versatile Dynamic Motions in Quadrupedal Robots》

《 Representation-Free Model Predictive Control for Dynamic Motions in Quadrupeds》

摘要--本文提出了一种新的无表示法控制无功补偿的模型预测控制(RF-MPC)框架四足机器人在三维空间中的动态运动(3维)空间。我们的公式直接代表了转动动力学--使用旋转矩阵的动力学，这将我们从问题中解放出来与使用欧拉角和四元数作为方向有关-注释表示法。具有基于变化的线性化方案和精心构造的代价函数，MPC控制律是转录成标准的二次规划形式。货币政策委员会控制器可以在四足动物上以250赫兹的实时速率运行机器人。包括周期性四足步态的实验结果和受控后空翻验证了我们的控制策略可以稳定涉及3-D奇点的动态运动行动。

部分代码展示：

function [A,B,D] = fcn_get_ABD_eta(Xt,Ut,p)
% linear dynamics for rotation
% evolution variable is eta

%% parameters
dt = p.Tmpc;

%% unpack
xop = reshape(Xt(1:3),[3,1]);
vop = reshape(Xt(4:6),[3,1]);
Rop = reshape(Xt(7:15),[3,3]);
wop = reshape(Xt(16:18),[3,1]);
pf34 = reshape(Xt(19:30),[3,4]);

%% constants for linear matrices
% [x,v,eta,w,constant]
[Cx_x,Cx_v,Cv_v,Cv_u,Cv_c] = eta_co_xv(Ut,dt,p.mass,p.g);
[CE_eta, CE_w, CE_c] = eta_co_R(Rop,wop,dt);
[Cw_x,Cw_eta,Cw_w, Cw_u, Cw_c] = eta_co_w(xop,Rop,wop,Ut,dt,p.J,pf34);


%% Assemble matrices
    A = [Cx_x, Cx_v, zeros(3,6);
         zeros(3), Cv_v, zeros(3,6);
         zeros(3,6),CE_eta,CE_w;
         Cw_x,zeros(3),Cw_eta,Cw_w];
    B = [zeros(3,12);
         Cv_u;
         zeros(3,12);
         Cw_u];
    D = [zeros(3,1);
         Cv_c;
         CE_c;
         Cw_c];

end

%% Core fcns for constant matrix
function [Cx_x,Cx_v,Cv_v,Cv_u,Cv_c] = eta_co_xv(fop,dt,mass,g)

Cx_x = eye(3);
Cx_v = eye(3) * dt;

Cv_v = eye(3);
Cv_u = dt/mass * [eye(3),eye(3),eye(3),eye(3)];
Cv_c = Cv_u * fop + [0;0;-g] * dt;

end

function [CE_eta, CE_w, CE_c] = eta_co_R(Rop,wop,dt)
% the input arguments are composed of variables at the operating point 
% and parameters

N = fcn_get_N;

%% debugged code
invN = pinv(N);

C_eta = kron(eye(3),Rop*hatMap(wop))*N + kron(eye(3),Rop)*fcn_get_D(wop);
C_w = kron(eye(3),Rop) * N;
C_c = vec(Rop*hatMap(wop)) - kron(eye(3),Rop)*N*wop;

CE_eta = eye(3) + invN * dt * kron(eye(3),Rop') * C_eta;
CE_w = invN * dt * kron(eye(3),Rop') * C_w;
CE_c = invN * dt * kron(eye(3),Rop') * C_c;

end

function [Cw_x,Cw_eta,Cw_w, Cw_u, Cw_c] = eta_co_w(xop,Rop,wop,fop,dt,J,pf)
% the input arguments are composed of variables at the operating point 
% and parameters

N = fcn_get_N;
r1 = pf(:,1) - xop;
r2 = pf(:,2) - xop;
r3 = pf(:,3) - xop;
r4 = pf(:,4) - xop;
Mop = [hatMap(r1) hatMap(r2) hatMap(r3) hatMap(r4)] * fop;

temp_J_w = hatMap(J*wop) - hatMap(wop) * J;
sum_fop = [eye(3),eye(3),eye(3),eye(3)] * fop;

Cx = Rop' * hatMap(sum_fop);
Ceta = fcn_get_F(Mop) * N - temp_J_w * hatMap(wop);
Cw = temp_J_w;
Cu = Rop' * [hatMap(r1),hatMap(r2),hatMap(r3),hatMap(r4)];
Cc = -hatMap(wop)*J*wop + Rop'*Mop - temp_J_w * wop - Cx*xop;

Cw_x = dt*(J\Cx);
Cw_eta = dt*(J\Ceta);
Cw_w = dt*(J\Cw) + eye(3);
Cw_u = dt*(J\Cu);
Cw_c = dt*(J\Cc);

end

%% Aux fcns
function F = fcn_get_F(k)

F = [k', zeros(1,3),zeros(1,3);...
     zeros(1,3),k',zeros(1,3);...
     zeros(1,3),zeros(1,3),k'];
end

function N = fcn_get_N

N = [0 0 0;...
     0 0 1;...
     0 -1 0;...
     0 0 -1;...
     0 0 0;...
     1 0 0;...
     0 1 0;...
     -1 0 0;...
     0 0 0];
end

function D = fcn_get_D(in)

d = in(1);
e = in(2);
f = in(3);
D = [0 0 0;
   e -d 0;
   f 0 -d;
   -e d 0;
   0 0 0;
   0 f -e;
   -f 0 d;
   0 -f e;
   0 0 0];
end

效果展示：

193号资源-源程序：SCI文档+程序动态四足机器人的自由模型预测控制-本人博客有解读资源-CSDN文库https://download.csdn.net/download/LIANG674027206/89795587 👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆下载资源链接👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆👆

《《《《《《《《更多资源还请持续关注本专栏》》》》》》》

论文与完整源程序_电网论文源程序的博客-CSDN博客https://blog.csdn.net/liang674027206/category_12531414.html

电网论文源程序-CSDN博客电网论文源程序擅长文章解读,论文与完整源程序,等方面的知识,电网论文源程序关注python,机器学习,计算机视觉,深度学习,神经网络,数据挖掘领域.https://blog.csdn.net/LIANG674027206?type=download