- What is the dimension and rank of the feature Jacobian, for a 8-dofs robot carrying a camera from where the Cartesian coordinates of 6 points are used?
The dimension of the interaction Jacobian is 12 × 6 12 \times 6 12×6,as it is a 8DOF robot,the dimenstion of a robot Jacobian is 6 × 8 6\times 8 6×8.
s ˙ = L s c W e J ( q ) q ˙ \dot \mathbf{s}=\mathbf{L_s}{}^c\mathbf{W}_e\mathbf{J}(q)\dot q s˙=LscWeJ(q)q˙
Thus, the dimension and rank of the feature Jacobian is 12 × 8 12 \times 8 12×8,rank6.
2.Why is the interaction matrix always assumed to be only estimated?
It depends on unknown 3D parameters
Centered moments lead to a more decoupled motion
For a specific feature s, how we can learn how many degree of freedom we can control?
IT depends on the shape of the feature,a point 2dof,a ball,3dof
we select m visual features to control m DOF,( k ≥ m , m ≤ 6 k \geq m, m \leq 6 k≥m,m≤6)
Give a definition of what we call Interaction Matrix and Image Jacobian. Explain why they are different.
We know the definition of the interaction matrix is about how visual features could vary when our camera moves. s ˙ = L s v c \dot s=\mathbf{L}_s\mathbf{v_c} s˙=Lsvc
about the definition of image Jacobian,which linearly relates changes in image features and changes in Cartesian coordinates or changes in joints .[1]It can be directly related to joint velocities as following.
s ˙ = J s q q ˙ = [ ∂ s 1 ∂ q 1 ⋯ ∂ s 1 ∂ q n ⋮ ⋱ ⋮ ∂ s m ∂ q 1 ⋯ ∂ s m ∂ q n ] q ˙ ; where J s q = J s r J r q = J \dot{\mathbf{s}}=\mathbf{J}_{s q} \dot{\mathbf{q}}=\left[\begin{array}{ccc} \frac{\partial s_{1}}{\partial q_{1}} & \cdots & \frac{\partial s_{1}}{\partial q_{n}} \\ \vdots & \ddots & \vdots \\ \frac{\partial s_{m}}{\partial q_{1}} & \cdots & \frac{\partial s_{m}}{\partial q n} \end{array}\right] \dot{\mathbf{q}} \quad ; \quad \text { where } \mathbf{J}_{s q}=\mathbf{J}_{s r} \mathbf{J}_{r q}=\mathbf{J} s˙=Jsqq˙=⎣⎢⎡∂q1∂s1⋮∂q1∂sm⋯⋱⋯∂qn∂s1⋮∂qn∂sm⎦⎥⎤