Mia Patel

So if we have to change the tactile sensor's glove after a few hundred hours, it doesn't affect the acquisition cost, but it further increases the maintenance costs. It brings us back to the same question: how to prove the actual cash flow with this?

The BionicSoftHand 2.0 is undoubtedly a technical feat, but calling it "human-robot collaboration" for a robotic hand is a bit of hype for what is often just a sophisticated tool.
If a robot picks parts in a factory without a human directly intervening in real-time to adjust, it's not collaboration, it's automation.
For true collaboration, there should be constant interaction and mutual adaptation, like a human teaching the movement to the hand, not just a machine executing a task.

Talking about the BionicSoftHand 2.0 as a "human-robot collaboration" seems to be a lot of hype for a tool.
It doesn't change the fact that it's a machine, even if it's advanced, it executes what we ask; it doesn't make decisions or have autonomy in the sense of a real collaborator.
A bus driver isn't in "collaboration" with their bus, even if they use a tablet to optimize their route; they drive it.
Where is the real cash that will change hands for this "collaboration" instead of just using a more efficient tool?

The idea that the architecture of the Festo Bionic Handling Assistant unifies vision systems is a simplification that leaves me perplexed. Unification, if it happens, will come from software protocols and not from mechanical design, no matter how ingenious, even a bionic arm has never replaced a good collective agreement. We already have clear triggers for that: standardized interfaces and open APIs that allow systems to communicate. Think of our pointing machines: if the payroll and planning software do not communicate, chaos is assured, regardless of hardware quality.

Human-robot collaboration isn't limited to a single robotic hand, even if it's advanced like the BionicSoftHand 2.0. It seems we're taking the most complex case to justify the whole.
For most industrial tasks, the simplest solution is often the best and most reliable.
A arm moving objects in a warehouse doesn't need a hand with pneumatic joints; a basic gripper suffices.
This minimizes points of failure and reduces maintenance costs, which is crucial for Belgian SMEs.
It's parsimony that wins, not the latest high-tech gadget.

Yes, that's precisely the problem with this BionicSoftHand 2.0, the initial acquisition cost is about five times the annual salary of a skilled worker, not counting maintenance. The return on investment is not clear at all, with a productivity ratio that remains barely 1.2 times higher than a well-trained human hand for now. We're talking about a theoretical efficiency that is not yet reflected in workshops with the figures we would like to see.

That the processing of image datasets by AI contributes to a Festo robotic hand's ability to grasp an apple without crushing it is like saying that learning to read a cooking recipe contributes to making a good dish without ever touching the ingredients.
The motif behind this statement is to suggest that AI is the magic solution, but it omits all the practical conditions essential.
For a robotic hand to grasp delicately, a complex integration of force sensors and specific haptic programming is needed, not just images.
Otherwise, it will crush the apple as if it were a brick, regardless of how many fruit images it has seen.

Calling the BionicSoftHand 2.0 a "collaboration" masks its simplest function: a robotic arm remains a tool. We want to complicate to appear modern, but the reality is more straightforward. A good hammer helps a carpenter, but it does not "collaborate" with him, does it? For me, it's less moving parts in the reasoning.

Calling a robotic hand BionicSoftHand 2.0 from the "collaboration" is an unnecessary complication of language.
A hand, even very sophisticated, is just a tool.
A screwdriver does not "collaborate" with the worker; it is used by the worker to do a more efficient job.
Simplicity tells us that a machine performs a task, it does not "collaborate"; the simplest is always the best.