In the last decade several efforts were devoted to model sediment‐particle transport in rivers as a stochastic process. Experimental observations are therefore needed to validate these models and to provide the correct probability distribution of selected stochastic variables. The kinematics of sand particles is investigated here using nonintrusive imaging to provide a statistical description of bedload transport under incipient motion conditions. In particular, we focus on the alternation between motion (particle steps) and rest regimes to quantify the probabilistic distribution of the particles waiting time, which is suggested by many studies to be responsible for anomalous diffusion. The probability distributions of the particle step time and step length, streamwise and spanwise velocities, acceleration, and waiting time are quantified experimentally. Results suggest that variables describing the particle motion regime are thin‐tailed distributed, whereas the waiting times exhibit a power law distribution. A specific class of waiting times during which the grain is observed to oscillate without a net displacement is classified as active and is analyzed separately from the other, so‐called deep waiting times. The experimental results, obtained under five different transport conditions, describe grain‐scale kinematics and dynamics at different wall shear stress. They provide both a benchmark data set for validating particle‐transport numerical simulation and critical input parameters for the stochastic modeling of bedload transport.