A very important aspect of humanoid robot design and construction is power supply and management. No matter how advanced we make a machine, if it is not able to operate with sufficient power intensity and duration, it will be of little use. This power must not only be sufficient, but must also be distributed optimally, and also be used efficiently within each component at all times.

Power supply and management in robots may be grouped under two main subtitles:

Efficiency of batteries

Power density and efficiency are primary factors that affect a robot’s overall operational capabilities and operational time, as well as the robot’s weight and size.

A battery’s internal resistance, charge/discharge rate, depth of discharge and operational temperatures all affect efficiency.

Power density is how quickly energy can be delivered per unit mass or volume of battery.

Power density should not be confused or grouped under efficiency. Efficiency basically tells us how well the battery gives us back, what we put in. To make an analogy, this is similar to fuel economy in cars. Power density on the other hand is how much instantaneous power a battery can deliver. This is similar to a race car being powerful but not so energy efficient.

In most humanoids, service robots, industrial robots, other types of mobile robots, Lithium-Ion (Li-ion) batteries are the most common, which have good balance between battery efficiency and power density. Lithium Polymer (LiPo) type of batteries are used more commonly in drones, robotic arms, race robots, where quick delivery (burst) of power may be needed often. We should also mention Lithium Iron Phosphate (LiFePO₄) batteries, which have lower energy density than Li-ion batteries but they are more stable and can have longer life cycles. They are mainly used in autonomous guided vehicles, industrial robots, collaborative robots.

Control algorithms to efficiently use and distribute available power

Battery control algorithms are used to optimize power efficiency, delivery and distribution in the components (such as motors, sensors, processors…) while balancing loads in the system and performing tasks as required. The tasks these algorithms do include the following:

• Regulate current and voltage to components, prevent voltage spikes, drops, overcurrent, short circuits, deep (zero or close to zero level) discharges, overheating, and ensure stable power delivery
• Scheduling and prioritizing tasks based on their importance and energy costs
• Deciding movement patterns
• Adjusting motion trajectories
• Reduce power or turn off less critical tasks
• Estimate state of battery and reassess power distribution based on expected tasks
• Use algorithms and reinforcement learning to continuously improve energy efficiency based on logged data of previously completed tasks and encountered environment

For specific components, these algorithms also optimize usage. For example they:

• Have the motor move with minimum required torque, to achieve the required motion
• During bipedal walking, optimize energy spent per step by adjusting gait patterns, step height and stride based on the terrain conditions, required speed
• Make use of passive dynamics, such as gravity or object’s inertia (momentum)
• Reduce or cut off power to non-critical sensors at a given moment
• Temporarily downgrade camera precision or resolution
• Generate energy during braking and slowing down of joints and direct this to batteries for recharging

Remember that, all of these are not decided in isolation, but coordinated collectively as part of a greater and multi layered control system. Also see our article here about control systems in humanoid robots.

By: A, Tuter

——————————————————

Terms of use:

Copying or republishing of any type of content from this website is not allowed without written permission from us. We make dated records of our posts, keep originals of our images.

The content in this website may contain errors, inaccuracies or be incomplete. User assumes all liability as a result of usage of content in our site. Also see our Terms page for more details.