Today was a pretty stressfull day. At nine o’clock in the morning stage one started with the Housekeeper tasks. We mainly went for Taking Out The Garbage and Storing Groceries. Because the teams got each an additional slot for both stage one categories, we also decided to try a run of ‘Serving the Breakfast’. Simultaneously to preparing the tasks, we had to calibrate TIAGos head in the morning and also an external camera we use. This went quite smooth. However the tasks already started at 9:00 a.m. in the morning and not all the games had enough preparation time — that’s why Lisa only mimed to prepare the breakfast by filling imaginary cereals into an imaginary bowl.
We had a perfect run of ‘Taking Out The Garbage’ — an actually useful task, right? I guess there are not many people that actually like this housekeeping task. To fulfill it, TIAGo went to the first approximate bin location and detected it using it’s laser scanner. We then try to find circular patterns in the laser scan and transform the detected points into map coordinates. To reliably grasp the trash bag out of the bin, an effort based grasping approach is used. As we don’t have a force torque sensor on the end-effector, we have to rely on the current of the motors which works also quite nice. So back to our approach on that: We lower TIAGo’s end-effector into the center of the bin until we get a certain threshold. That’s the signal to lift the trash bag. Because of the missing force-torque sensors we can hardly estimate if the trash slipped out of the end-effector or not — so we tried a lot to just ensure that we will grasp the bag reliably. This worked out twice and resulted in gaining 500 points for this task.
In the following, we did two rounds of Storing Groceries. However, as we could not porperly test this interesting manipulation task beforehand due to the late arrival of TIAGo, we sadly did not score any points. We will aim for that in tomorrows runs. The second part – the Partyhost tasks – we started with the ‘Receptionist’ task in which Lisa had to greet guests at the entrance door and introduce them to some other guests in the living room. “Introducing” meant stating the name and favorite drink of the person. The robot also had to point at the guest it was introducing and afterwards to an empty seat for them to sit. Lisa did a good job and got all possible points (no bonuses, though).
‘Serving Drinks’ is a game where the robot has to serve drinks to persons. People can be located randomly at a party location. They can choose between several drinks, but one of them is not available. By detecting the missing drink, it is possible to get bonus points. With a safe approach, we managed to make some points. Unfortunately, Lisa’s Kinect died while trying the game for a second time with a more advanced implementation. Tomorrow we’re going to run ‘Serving Drinks’ for another time after fixing some issues in an extended version.
As last game for today we also went for a round of ‘Find My Mates’, in which a person goes into an apartment and get a description of the persons in a specified room. We didn’t plan to test it here but the other ‘Partyhost’ tasks went that good and because of the mentioned extra round per scenario we just decided to spontaneously test this task which was just prepared on-site. Lisa did not that bad. She was able to find two persons in a room and described one of them correctly by their gender, pose (sitting/ standing) and the shirt color. The scores for today will follow soon - stay tuned!