1: \begin{abstract}

2: Constant biases associated to measured linear and angular velocities of a moving object can be estimated from  measurements of a static  scene  by  embedded brightness  and depth sensors. We propose here a  Lyapunov-based observer taking  advantage of the $SO(3)$-invariance of the  partial differential equations satisfied  by the measured  brightness and  depth fields. The resulting asymptotic observer is governed by a non-linear integro/partial differential system where the two  independent scalar variables indexing the pixels  live on $\SSS^2$.  The observer design  and analysis are  strongly simplified by  coordinate-free  differential calculus on $\SSS^2$ equipped with its natural   Riemannian structure.  The observer convergence is investigated under  $C^1$ regularity assumptions  on the object motion and its scene. It relies on Ascoli-Arzela theorem and pre-compactness of the observer trajectories. It is proved that the estimated biases  converge  towards the true ones, if and only if, the scene admits no cylindrical symmetry.  The observer design  can be adapted to realistic  sensors where brightness and depth data are only available on a subset of $\SSS^2$. Preliminary simulations with synthetic brightness  and depth images  (corrupted by noise around 10\%) indicate that such Lyapunov-based observers  should be  robust and  convergent  for much weaker regularity assumptions.

3: \end{abstract}