Adam Heins,Angela P. Schoellig
Abstract:We consider the nonprehensile object transportation task known as the waiter's problem - in which a robot must move an object balanced on a tray from one location to another - when the balanced object has uncertain inertial parameters. In contrast to existing approaches that completely ignore uncertainty in the inertia matrix or which only consider small parameter errors, we are interested in pushing the limits of the amount of inertial parameter uncertainty that can be handled. We first show how balancing constraints robust to inertial parameter uncertainty can be incorporated into a motion planning framework to balance objects while moving quickly. Next, we develop necessary conditions for the inertial parameters to be realizable on a bounding shape based on moment relaxations, allowing us to verify whether a trajectory will violate the balancing constraints for any realizable inertial parameters. Finally, we demonstrate our approach on a mobile manipulator in simulations and real hardware experiments: our proposed robust constraints consistently balance a 56 cm tall object with substantial inertial parameter uncertainty in the real world, while the baseline approaches drop the object while transporting it.
What problem does this paper attempt to address?
### Problems the paper attempts to solve
This paper aims to address the challenge of how robots handle the uncertainty of object inertial parameters when transporting objects. Specifically, it studies the "waiter's problem", that is, the robot needs to move an object smoothly from one position to another, and this object is placed on a tray only through frictional contact, and its inertial parameters (such as mass, center - of - mass position, and inertia matrix) are uncertain.
#### Main problem description
1. **Non - grasping object transportation**:
- The robot must move the object from one position to another without grasping it while maintaining the balance of the object on the tray.
2. **Inertial parameter uncertainty**:
- Inertial parameters such as the mass, center - of - mass position, and inertia matrix of the object are unknown or have large uncertainties. Existing methods usually ignore these uncertainties or only consider small - range parameter errors, while this paper is committed to dealing with large - range inertial parameter uncertainties.
3. **Balance constraints in fast motion**:
- During fast motion, ensure that the object does not slide or tip over relative to the tray. This requires the design of a robust motion planning framework that can adapt to large inertial parameter uncertainties.
4. **Verify the safety of the trajectory**:
- It is necessary to develop the necessary conditions to verify whether the planned trajectory will violate the balance constraints under any physically realizable inertial parameters, so as to ensure that the robot will not drop the object in actual operation.
#### Overview of solutions
- **Robust balance constraints**: By introducing robust balance constraints, the robot can successfully balance the object under large inertial parameter uncertainties.
- **Theoretical analysis**: Based on the moment relaxations technique, analyze how to ensure the satisfaction of balance constraints in the presence of uncertainties in the center - of - mass and inertia matrix.
- **Experimental verification**: Through simulation and actual hardware experiments, verify the effectiveness of the proposed robust constraint method, especially in dealing with high inertial parameter uncertainties and high object heights.
### Formula representation
The formulas involved in the paper are as follows:
1. **Spatial mass matrix**:
\[
\Xi = \begin{bmatrix}
I & -m c^\times \\
m c^\times & m
\end{bmatrix}
\]
where \( I \) is the inertia matrix of the object, \( m \) is the mass of the object, \( c \) is the center - of - mass position of the object, and \( c^\times \) is the skew - symmetric matrix of the center - of - mass position.
2. **Gravity - inertia torque**:
\[
w_{GI} = \Xi \eta - \text{ad}(\xi)^T \Xi \xi
\]
where \( \eta = [\dot{v}^T - g^T, \dot{\omega}^T]^T \), \( g \) is the gravitational acceleration, and \( \xi = [\omega^T, v^T]^T \) is the spatial velocity.
3. **Contact force constraint**:
\[
F_i f_i \leq 0
\]
where \( F_i \) is the linearized friction cone matrix of the \( i \) - th contact point, and \( f_i \) is the contact force of the \( i \) - th contact point.
4. **Robust balance constraint**:
\[
W_{GI}(\xi, \eta) \subseteq W_C
\]
where \( W_{GI}(\xi, \eta) = \{Y(\xi, \eta) \theta | \theta \in \Theta\} \), \( Y(\xi, \eta) \) is the regression matrix, and \( \Theta \) is the set of inertial parameters.
5. **Moment relaxation condition**:
\[
M_r(y) = \int_K b_d(r) b_d(r)^T d\rho(r)