Assessing lower limb motions is crucial for several applications in biomechanics and human-computer interaction. Nowadays, the displacement of lower limb joints is retrievable through piezoresistive strain gauges, magnetic sensors, and inertial measurement units, whereas non-invasive alternatives comprise surface electromyography (sEMG), brain-computer interfaces, and optical tracking. However, most prevailing technologies demand expensive, intricate hardware with multiple channels and rely on burdensome signal/video processing. Therefore, this paper presents an optical fiber sensor for assessing lower limb motions employing the force myography (FMG) technique. FMG predicts user movements and intentions from the radial pressures exerted by leg muscles, providing an intuitive force-level response and demanding fewer channels than the sEMG. Besides, fiber sensors are immune to variations of skin impedance and electromagnetic noise, circumventing the drawbacks of widespread force-sensing resistors and capacitive devices. In this work, a microbending optomechanical transducer converts muscle stimuli into light-intensity modulation, generating FMG waveforms that carry information about performed postures or movements. Firstly, experiments evaluated the average intensity versus knee joint displacements (0 to 120 • ) by attaching the optical transducers to the vastus intermedius and gastrocnemius/soleus muscles, achieving an average resolution of 5.4 • . Subsequently, we measured the sensor response to different movements to characterize and classify the performed actions according to their waveforms. The results demonstrate the sensor's feasibility for detecting lower limb motions based on a straightforward and non-invasive setup, motivating further applications in robotic exoskeletons for walking assistance and rehabilitation.