The advancement of the humanoid robotics field shows significant potential and progress, however humanoids still have many limitations. This is due to the complexities, cost, danger, and the functional limitations associated with the current state of this technology. This dissertation presents a new type of humanoid bipedal robot. It includes the invention, detailed design, software modeling, fabrication, testing, control, and motion analysis done on the robotic platform BALLU. Buoyancy Assisted Lightweight Legged Unit, or BALLU, is a safe, low cost, simple bipedal robot that uses a helium filled upper body, along with two thin carbon fiber legs. Due to the new architecture and buoyancy force, this robotic platform never falls down and is intrinsically stable. The main prototype version discussed is comparable in size to an adult human, and uses cable driven actuated knee joints for locomotion. The actuation components and the majority of the robot’s weight are located on the feet. Even though each leg has only one active degree of freedom, BALLU can walk forwards and backwards, jump, turn, climb and descend stairs, and go over obstacles and rough terrain. Understanding its non-intuitive dynamics, along with correct actuation and timing and control of the knee joints is the key factor for these types of locomotion.
Although BALLU’s leg architecture is inspired and is reminiscent of traditional humanoids and human legs, this novel approach to humanoid design behaves drastically differently and is not modeled using traditional robotic locomotion methods. This dissertation includes the foundational understanding and categorization of this platform's behavior. The platform's physics, control, and locomotion strategies are specifically analyzed and explained in detail. This robotic platform is not intended to replace traditional humanoid robots, rather to introduce new possibilities and present an ultra-safe, low-cost, light-weight alternative tuned for specific applications. The main contribution of this thesis is organization and classification of behavior understanding, and proposing novel unique methods of movement and locomotion for a buoyancy assisted biped robot.