The organization of structurally polarized microtubules into networks is critical for efficient cargo transport mediated by the molecular motors dynein and kinesin. The motility properties of molecular motors are best understood in simplified reconstituted systems using single microtubule filaments, as well as in cells with radial microtubule arrangements and axonal compartments with uniformly oriented microtubule arrays. However, it is not understood how active transport occurs in environments with more complicated cytoskeletal geometries, such as the mixed polarity microtubule arrays found in the dendrites of neurons. Here we focus on the plus-end directed kinesin-4 KIF21B motor that is associated with retrograde biased cargo movement in dendrites, despite the mixed polarity microtubule organization. How KIF21B achieves this net directional bias, as well as whether KIF21B is primarily responsible for retrograde directed motility is not known. To understand this, we examined KIF21B motility on mixed polarity microtubule arrays within in vitro systems of increasing complexity and in live neurons. In reconstituted systems with recombinant KIF21B and engineered dynamic antiparallel microtubule bundles or extracted mixed polarity dendritic microtubule arrays, the nucleotide-independent microtubule binding regions of KIF21B were shown to modulate microtubule dynamics and promote directional track switching. For analysis of KIF21B motility, existing methods to automate motor tracking were not ideal, and we developed a segmentation tool called Cega, to detect purified fluorescently labeled kinesin motors moving within a system with high background noise. Interestingly, KIF21B motors did not display the net directional bias along stabilized extracted dendritic microtubule arrays, as seen by KIF21B in live cells. This in combination with the dramatic stabilization of microtubule dynamics by KIF21B suggested that directional bias required microtubule remodeling by KIF21B motors, and thus would only be observed along native dynamic microtubule arrays. Unsurprisingly, KIF21B optogenetic recruitment to dendritic cargo induced net retrograde movement, and both native microtubule dynamics and the secondary microtubule binding regions of KIF21B were required to achieve this directional bias. These results suggest a mechanism where teams of cargo bound KIF21B motors coordinate nucleotide-sensitive and insensitive microtubule binding sites to regulate microtubule stability and promote track switching and ultimately achieve net retrograde movement along the mixed polarity microtubule arrays of dendrites.