Instrumented insoles could benefit locomotion research on healthy and clinical populations by providing data in natural settings outside of the laboratory. We designed a low-cost, instrumented insole with 8 pneumatic bladders to measure localized plantar pressure information. We collected gait data during treadmill walking at 1.0 m/s and 1.5 m/s and for sit-to-stand and stand-tosit tasks for 10 subjects. We estimated a common representation of ground kinetics (3-component force vector, 2-component center of pressure position vector, and a single-component torque vector) from the insole data. We trained an intertask neural network for each component of the kinetic data. For the walking tasks at 1.0 m/s and 1.5 m/s, the normalized root mean square error was between 3.1% and 12.9% and for the sit-to-stand and stand-to-sit tasks, the normalized root mean square error was between 3.3% and 21.3% Our findings suggest that the proposed low-cost, instrumented insoles could provide useful data about movement kinetics during real-world activities.