Geri Dön

Visual aided navigation and mapping for an autonomous rotorcraft

Başlık çevirisi mevcut değil.

  1. Tez No: 402650
  2. Yazar: MEHMET SUAT KAY
  3. Danışmanlar: PROF. CARLO L. BOTTASSO
  4. Tez Türü: Doktora
  5. Konular: Uçak Mühendisliği, Aircraft Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2013
  8. Dil: İngilizce
  9. Üniversite: Politechnico di Milano (Technical University of Milan)
  10. Enstitü: Yurtdışı Enstitü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 70

Özet

Özet yok.

Özet (Çeviri)

Autonomous rotorcraft in a confined environment that is preventing it to use a global localization system will end up with wrong estimates of velocity and position after a short time of dead reckoning initialization due to the inherent and external noise on the inertial sensors. Extracting corners from an image that is taken from an onboard camera and tracking them in subsequent images can be used as a velocity reference for state estimation algorithm. Scale ambiguity that is present in a monocular camera can be removed by using a fixed baseline stereo camera system. A measurement model using this setup with relative positional measurements is described in the first part of this thesis. The noise in the pixel measurements are back projected into three dimensional space and after calculation of the distance travelled during image acquisition interval, the weight of each feature point for the measurement update is characterized by the covariance matrix. This matrix obtained by algebraic combination of uncertainties in horizontal and vertical projection uncertainty of left and right cemeras. Extended Kalman Filtering with state equations for position in fixed frame,velocity in body frame and quaternion is used. Covariance representation for attitude representation is handled by a body fixed error approach described using Gibbs vector. Assertive simulation results are presented showing the improvement of velocity state vector. The second part discusses the mapping with known poses problem for a robotic vehicle. This is half of a more general problem called Simultaneous Localization and Mapping (SLAM). Occupancy grid mapping for vehicles moving on a two dimensional planar space is explained by many authors. An algorithm using this technique and utilizing dense disparity maps from a stereo camera is described for three dimensional mapping for a vehicle with six degrees of freedom. The algorithm generates random particles with probability distribution functions in accordance with the unertainty presented in the vehicle pose and vision sensors. The ratio of points collected in voxels to the total number of particles scattered for the measurement instant is used for a recursive Bayesian update. The validity of the approach is presented with simulation results.

Benzer Tezler

  1. Enhancing follow the gap method with memory aid and with prediction component

    Boşluğu takip et yönteminin hafıza desteği ile ve öngörü bileşeni ile geliştirilmesi

    EMRE CAN CONTARLI

    Yüksek Lisans

    İngilizce

    İngilizce

    2024

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Kontrol ve Otomasyon Mühendisliği Ana Bilim Dalı

    DOÇ. DR. VOLKAN SEZER

  2. Otonom taşıyıcı araçlar için yüksek çözünürlüklü harita güncelleme

    High definition map update for autonomous transfer vehicles

    MUHAMMED OĞUZ TAŞ

    Yüksek Lisans

    Türkçe

    Türkçe

    2018

    Elektrik ve Elektronik MühendisliğiEskişehir Osmangazi Üniversitesi

    Elektrik-Elektronik Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ HASAN SERHAN YAVUZ

    DOÇ. DR. AHMET YAZICI

  3. Harita mashup uygulamalarıyla zamana ve mekâna bağlı görsel analitik ortamı oluşturulması: Konya Kapalı Havzası örneği

    Developing a geo-temporal visual analytics environment with map mashups: A case study on Konya Closed Basin

    OSMAN SAMİ KIRTILOĞLU

    Doktora

    Türkçe

    Türkçe

    2016

    Jeodezi ve FotogrametriSelçuk Üniversitesi

    Harita Mühendisliği Ana Bilim Dalı

    PROF. DR. İBRAHİM ÖZTUĞ BİLDİRİCİ

  4. Human-space interaction in virtual landscapes

    Sanal peyzajlarda insan-mekan etkileşimi

    SUNA KORKMAZ

    Yüksek Lisans

    İngilizce

    İngilizce

    2023

    Mimarlıkİstanbul Teknik Üniversitesi

    Peyzaj Mimarlığı Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ IKHWAN KIM