This paper presents an initial prototype of an in-vivo robotic laparoscopic camera that features optimized illumination to address the problems, i.e. inferior lighting uniformity and low optical efficiency, in the state-of-the-art designs of in-vivo laparoscopic cameras. Benefiting from the transformable structure of the robotic camera, sufficient on-board space is created without sacrificing the camera's compactness to carry three dedicatedly-designed freeform optical lenses for achieving the optimized illumination requirements. Designing miniature freeform optical lenses for extended light sources, such as LEDs, is a very challenging task which usually involves solving a non-standard Monge-Ampere equation. In this paper, we approach the illumination optical design based on a ray-mapping method that is governed by a standard Monge-Ampere equation. We propose an effective numerical solver to compute the ray-mapping solution for constructing freeform lens surfaces. Experimental results prove the predicted performance of the illumination system design: greater than 97% illuminance uniformity, greater than 80% optical efficiency, and greater than 14,323 lx illuminance on a target plane with a distance of 100 mm. The effectiveness of this prototype is also experimentally verified by performing a suturing task in a simulated abdomen.

RGB-D SLAM systems based on 3D Gaussian Splatting (3DGS) often suffer from map degradation caused by diminishing historical supervision, noisy depth observations, and local-window optimization during online reconstruction. To address these issues, we propose DUDG-SLAM, a Dynamic Replay and Depth-Uncertainty-Guided Gaussian SLAM framework. The proposed method selectively replays historical keyframes according to reconstruction error, forgetting degree, and Gaussian visibility, thereby restoring supervision in previously reconstructed regions. In addition, RGB-D sensor depth is retained as the primary geometric constraint, while scale-aligned Depth Pro predictions are introduced only in regions with missing or unreliable sensor depth. A pixel-wise uncertainty-weighted log-depth loss is further designed to reduce the influence of noisy depth observations and unreliable depth boundaries. Experiments on Replica and TUM RGB-D show that DUDG-SLAM improves rendering quality, trajectory accuracy, and reconstruction robustness. Compared with the baseline, DUDG-SLAM consistently improves PSNR and SSIM while reducing LPIPS and ATE.